更新赛题三演示视频

新增 提交材料/子赛题二-Skills/README.md，逐条对照子赛题二要求：
- 遵循 gitlink-cli Skills 规范（skills/README.md）
- 每个 Skill 含 SKILL.md + 使用示例（examples/）+ REFERENCE.md
- Claude Code 平台验证通过 + 兼容
- 演示录屏/截图 + 演示输入提示词
并列出 23 个 Skill 位置（skills/ 19 + workflow/skills 4）与智能分析型/命令操作型分类。

.devops/构建部署Showcase.yml

@@ -21,7 +21,7 @@ workflow:
       ssh_port: '"22"'
       ssh_user: '"root"'
       ssh_cmd: >-
-        "cd /opt/gitlink-cli && git fetch origin && git reset --hard origin/master && docker build -f showcase/Dockerfile -t gitlink-cli-showcase . && docker stop gitlink-cli-showcase || true && docker rm gitlink-cli-showcase || true && docker run -d -p 9090:9090 --name gitlink-cli-showcase --restart unless-stopped -e GITLINK_TOKEN='cookie:autologin_trustie=56c6d2b4378588465c97d48679eed7bd2495ff1a' gitlink-cli-showcase"
+        "git config --global --add safe.directory /opt/gitlink-cli && cd /opt/gitlink-cli && git fetch origin && git reset --hard origin/master && DOCKER_BUILDKIT=1 docker build -f showcase/Dockerfile -t gitlink-cli-showcase . && docker stop gitlink-cli-showcase || true && docker rm gitlink-cli-showcase || true && docker run -d -p 9090:9090 --name gitlink-cli-showcase --restart unless-stopped -e GITLINK_TOKEN='cookie:autologin_trustie=56c6d2b4378588465c97d48679eed7bd2495ff1a' gitlink-cli-showcase"
     needs:
       - start
   - ref: end

.gitignore

@@ -0,0 +1,37 @@
+# Binaries
+*.exe
+*.exe~
+*.dll
+*.so
+*.dylib
+bin/
+dist/
+
+# Showcase build outputs（由 Dockerfile / deploy 脚本编译，不入库）
+showcase/showcase-linux
+showcase/showcase-server-linux
+showcase/showcase-server
+showcase/showcase
+
+# Test binary
+*.test
+
+# Output of go coverage
+*.out
+
+# Go workspace
+go.work
+
+# IDE
+.idea/
+.vscode/
+*.swp
+*.swo
+
+# OS
+.DS_Store
+Thumbs.db
+
+# Env
+.env
+.env.local

PROJECT_STATUS.md

@@ -1,279 +0,0 @@
-# GitLink-CLI 项目状态总结
-
-> 最后更新：2026-06-03
-> 仓库：`D:\自用\self\word\大三下\软件演化\gitlink-cli`
-> 远程：`https://gitlink.org.cn/chroe/gitlink-cli.git`
-
----
-
-## 一、课程任务背景
-
-**课程**：《软件演化与运维》课程实践
-**进阶任务**：GitLink 智能化能力提升项目
-**当前阶段**：子任务一 — 增加和完善 GitLink-CLI 能力（50%，截止 6月4日）
-
-### 子任务一交付要求
-- 向 gitlink-cli 主仓库提交 PR（可多个）
-- 每个 PR 包含：功能代码 + 单元测试 + 命令帮助文档更新
-- 提供变更说明文档
-- 撰写《软件分析及建模报告》、《新需求构思报告》、《变更影响分析及测试报告》
-
----
-
-## 二、技术栈与架构
-
-- **语言**：Go 1.26.1
-- **CLI 框架**：spf13/cobra
-- **密钥存储**：zalando/go-keyring（OS keychain + 文件 fallback）
-- **配置**：gopkg.in/yaml.v3，存放于 `~/.config/gitlink-cli/config.yaml`
-- **三层命令架构**：
-  1. **基础命令**（cmd/）：`auth`, `api`, `config`, `version`
-  2. **Shortcut 命令**（shortcuts/）：`repo +list`, `issue +create` 等 94 个命令
-  3. **Raw API**：`api GET /path`
-
-### Shortcut 模块开发模式
-
-每个模块位于 `shortcuts/<name>/<name>.go`，结构固定：
-
-```go
-package <name>
-
-import "github.com/gitlink-org/gitlink-cli/shortcuts/common"
-
-func Shortcuts() []*common.Shortcut {
-    return []*common.Shortcut{
-        {
-            Name:        "list",
-            Description: "Description here",
-            Flags: []common.Flag{
-                {Name: "page", Short: "p", Usage: "Page number", Default: "1"},
-            },
-            Run: func(ctx *common.RuntimeContext) error {
-                if err := ctx.ResolveOwnerRepo(); err != nil { return err }
-                env, err := ctx.CallAPI("GET", ctx.RepoPath()+"/something", nil)
-                if err != nil { return err }
-                return ctx.Output(env)
-            },
-        },
-    }
-}
-```
-
-注册到 `shortcuts/register.go`：
-```go
-import "github.com/gitlink-org/gitlink-cli/shortcuts/<name>"
-// 在 RegisterAll() 中添加:
-// "<name>": <name>.Shortcuts(),
-// descriptions["<name>"] = "Description",
-```
-
-### 单元测试模式（httptest mock）
-
-```go
-func TestXxx(t *testing.T) {
-    server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-        // 校验 method/path/query，返回 mock JSON
-    }))
-    defer server.Close()
-
-    ctx := &common.RuntimeContext{
-        Client: &client.Client{HTTP: server.Client(), BaseURL: server.URL},
-        Owner: "owner", Repo: "repo", Format: "json", Args: args,
-    }
-    err := findShortcut(t, "list").Run(ctx)
-    // 断言
-}
-```
-
----
-
-## 三、已实现的 18 个 Shortcut 模块（94 个命令）
-
-| 模块 | 命令数 | 文件 | 测试 |
-|------|--------|------|------|
-| repo | 9（含 batch-delete/fork） | shortcuts/repo/ | ✅ |
-| issue | 12（含 batch-close/assign/label/milestone） | shortcuts/issue/ | ✅ |
-| pr | 9 | shortcuts/pr/ | ✅ |
-| release | 5 | shortcuts/release/ | ✅ |
-| branch | 5 | shortcuts/branch/ | ✅ |
-| ci | 4 | shortcuts/ci/ | ✅ |
-| commit | 4 | shortcuts/commit/ | ✅ |
-| file | 5 | shortcuts/file/ | ✅ |
-| member | 4 | shortcuts/member/ | ✅ |
-| star | 3 | shortcuts/star/ | ✅ |
-| watch | 3 | shortcuts/watch/ | ✅ |
-| label | 4 | shortcuts/label/ | ✅ |
-| milestone | 6 | shortcuts/milestone/ | ✅ |
-| webhook | 6 | shortcuts/webhook/ | ✅ |
-| org | 5（含 batch-invite） | shortcuts/org/ | ✅ |
-| search | 3 | shortcuts/search/ | ✅ |
-| user | 2 | shortcuts/user/ | ✅ |
-| wiki 🆕 | 5 | shortcuts/wiki/ | ✅ 12个测试 |
-
-### 批量操作汇总（7个）
-
-| 批量命令 | 所属模块 | 输入方式 | 预览模式 |
-|----------|----------|----------|----------|
-| batch-close | issue | --numbers 逗号 / --from CSV | --dry-run |
-| batch-assign | issue | 同上 | --dry-run |
-| batch-label | issue | 同上 | --dry-run |
-| batch-milestone | issue | 同上 | --dry-run |
-| batch-delete | repo | --repos owner/repo 列表 / --from CSV | --dry-run |
-| batch-fork | repo | 同上 | --dry-run |
-| batch-invite | org | --users 用户名列表 / --from CSV | --dry-run |
-
----
-
-## 四、Wiki 模块开发记录
-
-### Wiki API 关键发现
-
-| 发现 | 说明 |
-|------|------|
-| API 域名 | `gateway.gitlink.org.cn` 而非 `www.gitlink.org.cn` |
-| API 路径 | `/wiki/open/...`（有 `/open/` 中间段） |
-| 无 `.json` 后缀 | Gateway 不接受 `.json`，需要 `client.DoRaw()` |
-| Sidebar 名称 | `_Sidebar`（大写 S），非小写 `_sidebar` |
-| projectId 字段 | 仓库返回 `project_id` 而非 `id` |
-| create/update 需要 `title` | API 必填字段 |
-| delete 异步重建 sidebar | 需延迟 2 秒后再清理 sidebar |
-| `data.data` 格式 | `DoRaw` 返回 dict 而非 string，解析需兼容 |
-
-### delete 命令的 sidebar 自动清理流程
-1. 删除 wiki 页面（DELETE）
-2. 等待 2 秒（让 GitLink 完成异步 sidebar 重建）
-3. 获取 `_Sidebar` 内容
-4. 移除已删页面的 `[[PageName]]` 链接
-5. 更新 `_Sidebar`
-
----
-
-## 五、开发踩坑记录
-
-### 修改 CLI 代码后必须同时重建 showcase
-showcase 通过 `findCLIBinary()` 在运行时查找 `gitlink-cli.exe`。每次修改 CLI 代码后需要：
-```bash
-cd gitlink-cli
-go build -o gitlink-cli.exe .          # 重建 CLI
-cd showcase && go build -o showcase.exe .  # 重建 showcase（更新嵌入的 HTML）
-# 然后重启 showcase.exe
-```
-如果只重建 showcase 不重建 CLI，showcase 用的还是旧版二进制。
-
-### Showcase 参数传递坑
-showcase 后端曾用 `strings.Split(args, " ")` 按空格切割参数，导致 `"Hello Wiki!"` 被截断为 `"Hello"`。
-修复：前端给含空格的值加引号，后端用支持引号解析的 `parseShellArgs()` 替代简单 Split。
-
-### Showcase 布尔参数坑（已修复）
-Cobra 的 BoolP 标志：`--dry-run false`（空格分隔）会被解析为 `--dry-run=true` + 多余参数 "false"。
-所以批量操作 `--dry-run false` 实际上是在预览模式下运行，不会真正执行。
-修复：布尔参数改用 `<select>` 下拉框，JS 只在选"是"时才传 `--flag`，选"否"时不传。
-
-### Showcase 重复参数坑（已修复）
-服务器后端自动添加 `--owner chroe --repo gitlink-cli`。如果前端卡片也传 `--owner`/`--repo`，
-会导致参数重复。修复：watch +watchers 和 star +stars 移除了多余的 owner/repo 输入框。
-
-### Showcase null 显示（已修复）
-CLI 命令返回空数据时，前端显示 "null"。修复：JS 层对 null/空对象/空字符串统一显示"操作完成，无返回数据"。
-
----
-
-## 六、CI/CD 流水线
-
-### 部署架构
-```
-代码 push → GitLink 触发流水线 → SSH 到服务器 → git fetch + reset → docker build → 重启容器
-```
-
-### 流水线文件
-- `.devops/构建部署Showcase.yml` — GitLink DevOps 流水线（push 自动触发）
-
-### 流水线关键配置
-- 使用 `git fetch + git reset --hard origin/master` 避免 git pull 的本地修改冲突
-- 使用 `docker build --no-cache` 确保每次用最新代码构建
-- 容器启动时注入 `-e GITLINK_TOKEN=cookie:autologin_trustie=...` 解决认证问题
-- GitLink 密钥管理：`deploy_server.server_password` = 服务器密码
-
-### 服务器信息
-- **IP**: 118.31.4.168
-- **用户**: root
-- **部署路径**: /opt/gitlink-cli
-- **容器名**: gitlink-cli-showcase
-- **端口**: 9090
-- **访问地址**: http://118.31.4.168:9090
-- **Showcase 展示模块数**: 12 个模块卡片（milestone/webhook/label/commit/wiki/file/member/watch/star/issue批量/repo批量/org批量）
-
-### Docker 构建注意
-Dockerfile 已配置 `ENV GOPROXY=https://goproxy.cn,direct`，解决国内网络 Go 模块下载问题。
-服务器 Docker daemon 已配置国内镜像加速器（`/etc/docker/daemon.json`）。
-
----
-
-## 七、已完成工作
-
-### ✅ 子任务一完成项
-- [x] 基础命令框架（auth/api/config/version）
-- [x] 18 个 Shortcut 模块（94 个命令），其中 wiki 为新增
-- [x] 7 个批量操作命令（batch-close/assign/label/milestone/delete/fork/invite）
-- [x] 12 个 AI Agent Skills
-- [x] Showcase Dashboard 在线展示（12 个模块卡片，支持真实运行）
-- [x] GitHub CI/CD（release + npm publish + FreeBSD 支持）
-- [x] GitLink DevOps 流水线（push 自动部署，已配置密钥，正常运行）
-- [x] 单元测试（wiki 12个 + 其他模块 47+）
-- [x] client.DoRaw 方法（不带 .json 后缀的原始 API 调用）
-- [x] Showcase 布尔参数/重复参数/null显示 修复
-
-### 待写文档 📄
-- [ ] 《软件分析及建模报告》
-- [ ] 《新需求构思报告》
-- [ ] 《变更影响分析及测试报告》
-
----
-
-## 八、小组分工
-
-| 角色 | 负责人 | 负责内容 |
-|------|--------|----------|
-| 组长 A | — | commit(4命令)、milestone(6)、webhook(6)、label(4)、Showcase |
-| 同学 B | — | file(5)、member(4)、watch(3)、star(3) |
-| 同学 C | — | 批量操作增强(batch-assign/label/milestone/delete/fork/invite)、table 输出优化、测试覆盖、FreeBSD构建 |
-| 全员 | — | wiki(5命令)、流水线部署 |
-
----
-
-## 九、快速开始
-
-```bash
-# 构建
-cd gitlink-cli && make build
-
-# 运行测试
-go test ./shortcuts/... -v
-
-# 登录
-./gitlink-cli auth login
-
-# 使用示例
-./gitlink-cli repo +list --owner gitlink
-./gitlink-cli issue +list --owner gitlink --repo gitlink-cli
-./gitlink-cli wiki +list --owner chroe --repo gitlink_help_center
-```
-
----
-
-## 十、当前进度 & 下一步
-
-### 已完成 ✅
-1. 全部 18 个 Shortcut 模块 + 94 个命令
-2. 7 个批量操作命令
-3. Showcase 展示页（12 个模块，可在线运行）
-4. GitLink DevOps 流水线（push 自动部署）
-5. 单元测试 47+ 个
-6. Showcase 各种 bug 修复（布尔参数、null 显示、重复参数、member ID 提示）
-
-### 下一步 📋
-1. 向 gitlink-cli 主仓库提交 PR
-2. 撰写《软件分析及建模报告》
-3. 撰写《新需求构思报告》
-4. 撰写《变更影响分析及测试报告》
-5. 变更说明文档

README.md

@@ -1,430 +1,183 @@
-# gitlink-cli
+# GitLink CLI · 智能化能力提升
 
-[![GitLink](https://img.shields.io/badge/GitLink-Gitlink%2Fgitlink--cli-green)](https://www.gitlink.org.cn/Gitlink/gitlink-cli)
-[![License](https://img.shields.io/badge/License-MulanPSL--2.0-blue.svg)](https://license.coscl.org.cn/MulanPSL2)
-[![Go Version](https://img.shields.io/badge/Go-1.26%2B-blue.svg)](https://golang.org)
-[![npm version](https://img.shields.io/npm/v/@gitlink-ai/cli.svg)](https://www.npmjs.com/package/@gitlink-ai/cli)
+> 让 AI 智能体和开发者都能在终端高效操控 GitLink 平台，把仓库管理、Issue/PR 协作、CI/CD 变成可自动化、可复现的流程。
 
-The official [GitLink](https://www.gitlink.org.cn) CLI tool — built for humans and AI Agents. Supports **macOS, Linux, and Windows**. Covers repository management, issue tracking, pull requests, CI/CD, and AI-powered workflows, with 40+ commands and 12 AI Agent [Skills](./skills/).
+**第八届 CCF 开源创新大赛 · track1 GitLink-CLI 贡献赛**参赛作品。
 
-**[中文文档](./README.zh-CN.md)**
+---
 
-[Install](#installation--quick-start) · [AI Agent Skills](#ai-agent-skills) · [Auth](#configure--use) · [Commands](#usage-examples) · [Contributing](#related-projects)
+## 项目简介
 
-## Why gitlink-cli?
+随着 Claude Code、Cursor 等 AI 编程智能体兴起，开发者正从"手动操作平台"转向"Agent 驱动开发"。但 GitLink 平台的能力长期锁在网页 GUI 里——AI 看不到、调不动，开发者也得在终端和浏览器间反复切换。
 
-- **Agent-Native Design** — 12 structured [Skills](./skills/) out of the box, compatible with Claude Code, OpenClaw, and other AI platforms — Agents can operate GitLink with zero extra setup
-- **Wide Coverage** — Repository, Issue, PR, Branch, Release, CI, Org, Search, User — all core domains covered
-- **AI-Friendly & Optimized** — Every command is tested with real Agents, featuring concise parameters, smart defaults, and structured output
-- **Cross-Platform** — Runs on macOS, Linux, and Windows (x64/arm64), install via `npm install -g @gitlink-ai/cli` in one command, binary auto-downloaded
-- **Open Source, Zero Barriers** — MulanPSL-2.0 license, ready to use, just `npm install`
-- **Up and Running in 3 Minutes** — Interactive login or `GITLINK_TOKEN` env var, from install to first API call in just 3 steps
-- **Secure & Controllable** — OS-native keychain credential storage, `GITLINK_TOKEN` env var for CI/CD & non-interactive environments, auto git remote context resolution
-- **Three-Layer Architecture** — Shortcuts (human & AI friendly) → Raw API (full coverage) → Config (configuration management)
+本项目围绕 [gitlink-cli](https://gitlink.org.cn/gitlink/gitlink-cli) 做了四件事：
 
-## Features
+1. **把平台能力命令化**——新增 19 个命令模块、110 条命令，让每类操作都能在终端一行命令完成
+2. **给 AI 写使用说明书**——23 个 Skill，让 Claude Code 等智能体直接会操控 GitLink
+3. **串成端到端工作流**——自研工作流引擎，一行命令把多个 Skill 串成自动化流水线
+4. **延伸到科研场景**——科研项目洞悉、FAIR 合规检查、贡献者画像
 
-| Category | Capabilities |
-|----------|-------------|
-| 📦 Repo | List, create, fork, delete repositories, view repo info |
-| 🐛 Issue | Create, update, close, batch close, comment on issues |
-| 🔀 PR | Create, merge, review pull requests, view changed files |
-| 🌿 Branch | Create, delete, list, protect, unprotect branches |
-| 🏷️ Release | Create, view, delete releases |
-| 🏢 Org | Manage organizations, members, teams |
-| 🔧 CI | View builds, logs, CI/CD operations |
-| 🔍 Search | Search repositories, users |
-| 👤 User | View user profiles and info |
-| 📋 PM | Sprint management, kanban boards, weekly reports |
-| 🤖 Workflow | AI-powered issue triage, PR review, release notes |
+| 指标 | 数值 |
+|------|------|
+| 命令模块 | 19 个（110 条命令） |
+| 批量操作 | 11 条（全部支持 `--dry-run` 预演） |
+| AI Agent Skill | 23 个 |
+| 预置工作流 | 5 个 |
+| 单元测试 | 187 个 |
+| 交叉编译平台 | 8 个（Win/Linux/macOS/FreeBSD × amd64/arm64） |
 
-## Installation & Quick Start
+---
 
-### Requirements
+## 四大子赛题成果
 
-- Node.js 14+ (`npm`/`npx`) — for npm installation
-- Supported platforms: macOS, Linux, Windows (x64/arm64)
-- Go 1.26+ — only required for building from source
+### 子赛题一 · 增加 CLI 能力
+新增 19 个 Shortcut 模块（仓库/Issue/PR/Release/CI/Webhook/标签/里程碑/Wiki/工作流…），11 条批量操作，修复跨平台兼容性（Windows/Linux/macOS/FreeBSD）。已向主仓库提交 PR。
+- 代码：[`shortcuts/`](shortcuts/)、[`internal/`](internal/)、[`cmd/`](cmd/)
+- 提交材料：[`提交材料/子赛题一-CLI能力/`](提交材料/子赛题一-CLI能力/)
 
-### Quick Start (Human Users)
+### 子赛题二 · 编写 Skills
+23 个 Skill（命令操作类 + 智能分析型），覆盖健康度报告、Release Notes 生成、Issue 智能分拣、智能代码审查、许可证合规、科研辅助等。一键 `setup-skills.sh` 装进 Claude Code 即可用 `/gitlink-xxx` 调用。
+- 代码：[`skills/`](skills/)
+- 演示录屏：[`提交材料/子赛题二-Skills/demos/`](提交材料/子赛题二-Skills/demos/)
 
-> **Note for AI assistants:** If you are an AI Agent helping the user with installation, jump directly to [Quick Start (AI Agent)](#quick-start-ai-agent), which contains all the steps you need to complete.
+### 子赛题三 · 端到端工作流
+自研工作流引擎（`shortcuts/workflow/`），AI/规则双模式，5 个预置工作流（代码质量看门人/社区运营自动化/贡献者成长/多仓库协同/项目一键初始化），支持手动/轮询/定时触发，systemd 7×24 自动运行。
+- 代码：[`shortcuts/workflow/`](shortcuts/workflow/)
+- 提交材料：[`提交材料/子赛题三-工作流/`](提交材料/子赛题三-工作流/)
 
-#### Install
+### 子赛题四 · 辅助科研
+gitlink-spark（科研软件 X 光）、gitlink-research-fair（FAIR 合规检查）、gitlink-contributor-ranking（贡献者画像），面向科研项目洞悉、合规校验、协作匹配。
+- 代码：[`skills/gitlink-spark`](skills/gitlink-spark)、[`skills/gitlink-research-fair`](skills/gitlink-research-fair)
+- 提交材料：[`提交材料/子赛题四-科研/`](提交材料/子赛题四-科研/)
 
-**From npm (recommended):**
+> 📁 所有变更说明、演示录屏/截图集中在 [`提交材料/`](提交材料/) 目录。
+
+---
+
+## 快速开始
+
+### 安装
 
 ```bash
-# One command: installs CLI binary + all 12 AI Agent Skills
+# 方式一：go install（需 Go 1.26+）
+go install github.com/gitlink-org/gitlink-cli@latest
+
+# 方式二：npm
 npm install -g @gitlink-ai/cli
+
+# 方式三：下载预编译包（8 平台）
+# 见 Release 页面
 ```
 
-The binary is auto-downloaded for your platform during `postinstall`. No extra steps needed.
-
-**From source:**
-
-Requires Go 1.26+.
+### 认证
 
 ```bash
-git clone https://www.gitlink.org.cn/Gitlink/gitlink-cli.git
-cd gitlink-cli
-make install
+gitlink-cli auth login    # 浏览器登录，cookie 自动保存
+gitlink-cli auth status   # 查看登录状态
 ```
 
-> **Windows users:** Run `npm install -g @gitlink-ai/cli` in PowerShell or CMD. For building from source, use `go install .` instead of `make install`.
-
-#### Configure & Use
+### 使用示例
 
 ```bash
-# 1. Configure (one-time, interactive guided setup)
-gitlink-cli config init
+# 列出仓库的 Issue
+gitlink-cli issue +list --owner gitlink --repo gitlink-cli
 
-# 2. Log in (choose one)
-gitlink-cli auth login            # Username/password (recommended)
-gitlink-cli auth login --token    # Or paste a private token
-export GITLINK_TOKEN="your-token" # Or set env var (for CI/CD, non-interactive environments)
+# 批量关闭（先预演）
+gitlink-cli issue +batch-close --numbers 1,2,3 --dry-run
 
-# 3. Start using
-gitlink-cli repo +list
+# 在 Claude Code 里一句话调用 Skill
+/gitlink-triage 帮我分拣所有未分类的 Issue
 ```
 
-### Quick Start (AI Agent)
-
-> The following steps are for AI Agents. Some steps require the user to complete actions in a browser.
-
-**Step 1 — Install**
+### 一键启用 Skills（Claude Code）
 
 ```bash
-# One command: CLI binary + all Skills auto-installed
-npm install -g @gitlink-ai/cli
+bash scripts/setup-skills.sh   # 把 skills/ 链接到 ~/.claude/skills/
+# 之后在 Claude Code 里就能 /gitlink-xxx 调用
 ```
 
-**Step 2 — Configure**
+### 在线演示
 
-```bash
-gitlink-cli config init
+Showcase Dashboard：<http://118.31.4.168:9090> —— 浏览器里点卡片就能跑命令，无需安装。
+
+---
+
+## 架构说明
+
+```
+┌──────────────────────────────────────────────────────────┐
+│ 入口层  cmd/        main.go → root.go (cobra root)       │
+│   4 个基础命令: auth / api / config / version            │
+│   全局 flag: --owner/--repo/--format/--debug             │
+└───────────────┬──────────────────────────────────────────┘
+                │ shortcuts.RegisterAll(rootCmd)
+┌───────────────▼──────────────────────────────────────────┐
+│ 业务层  shortcuts/   19 组 × 100+ 命令 + 工作流引擎       │
+│   register.go   命令注册中枢                              │
+│   common/       RuntimeContext 业务统一上下文             │
+│   workflow/     工作流引擎（AI/规则双模式 + 5 预置流）     │
+└───────────────┬──────────────────────────────────────────┘
+                │ RuntimeContext.CallAPI / Output
+┌───────────────▼──────────────────────────────────────────┐
+│ 基础层  internal/   Go 内部包，不对外暴露                 │
+│   client/   HTTP 客户端 + .json 后缀 + 错误检测            │
+│   auth/     Transport 自动注入 Cookie/access_token         │
+│   config/   YAML 配置 + 跨平台路径                        │
+│   output/   Envelope(ok/data/meta) + json/yaml/table      │
+└──────────────────────────────────────────────────────────┘
+
+独立的 AI 层  skills/   23 个 SKILL.md，面向 AI Agent 的自然语言说明书
+                          （非编译期依赖，指导 Agent 调用编译好的二进制）
 ```
 
-**Step 3 — Login**
+**核心设计**：业务命令只依赖 `RuntimeContext`，绝不直接接触 cobra/HTTP 细节——新增模块只挂一个 `shortcuts/` 子包，不动入口与基础层。这是 19 模块 100+ 命令仍保持可测试性的根本。
 
-For interactive environments:
-```bash
-gitlink-cli auth login
-```
+---
 
-For non-interactive environments (CI/CD, Trae sandbox, MCP, etc.):
-```bash
-export GITLINK_TOKEN="your-private-token"
-```
-
-> To get a private token, go to GitLink web → Settings → Private Tokens.
-
-**Step 4 — Verify**
-
-```bash
-gitlink-cli user +me
-```
-
-## Usage Examples
-
-### Repository Operations
-
-```bash
-# List repositories
-gitlink-cli repo +list
-
-# View repository info
-gitlink-cli repo +info --owner Gitlink --repo forgeplus
-
-# Create a repository
-gitlink-cli repo +create -n my-project -d "Project description"
-
-# Fork a repository
-gitlink-cli repo +fork --owner Gitlink --repo forgeplus
-```
-
-### Issue Management
-
-```bash
-# List issues
-gitlink-cli issue +list --owner Gitlink --repo forgeplus
-
-# Create an issue
-gitlink-cli issue +create --owner Gitlink --repo forgeplus -t "Bug: Login failed" -b "Steps to reproduce..."
-
-# View an issue
-gitlink-cli issue +view --owner Gitlink --repo forgeplus -i 123
-
-# Close an issue
-gitlink-cli issue +close --owner Gitlink --repo forgeplus -i 123
-
-# Preview batch close without changing data
-gitlink-cli issue +batch-close --owner Gitlink --repo forgeplus --numbers 123,124 --dry-run
-
-# Batch close issues from a CSV file
-gitlink-cli issue +batch-close --owner Gitlink --repo forgeplus --from issues.csv
-
-# Add a comment
-gitlink-cli issue +comment --owner Gitlink --repo forgeplus -i 123 -b "Fixed"
-```
-
-### Pull Requests
-
-```bash
-# List PRs
-gitlink-cli pr +list --owner Gitlink --repo forgeplus
-
-# Create a PR (same-repo branch)
-gitlink-cli pr +create --owner Gitlink --repo forgeplus -t "feat: Search feature" --head feature/search --base master
-
-# Create a PR (from a fork)
-gitlink-cli pr +create --owner Gitlink --repo forgeplus -t "feat: New feature" --head your_username/forgeplus:feature/my-feature --base master
-
-# View a PR
-gitlink-cli pr +view --owner Gitlink --repo forgeplus -i 42
-
-# Merge a PR
-gitlink-cli pr +merge --owner Gitlink --repo forgeplus -i 42
-
-# View changed files
-gitlink-cli pr +files --owner Gitlink --repo forgeplus -i 42
-```
-
-### Branch Management
-
-```bash
-# List branches
-gitlink-cli branch +list --owner Gitlink --repo forgeplus
-
-# Create a branch
-gitlink-cli branch +create --name feature/new-feature
-
-# Delete a branch
-gitlink-cli branch +delete --name feature/old-feature
-
-# Protect a branch
-gitlink-cli branch +protect --name main
-
-# Remove branch protection
-gitlink-cli branch +unprotect --name main
-```
-
-### Release Management
-
-```bash
-# List releases
-gitlink-cli release +list --owner Gitlink --repo forgeplus
-
-# Create a release
-gitlink-cli release +create --owner Gitlink --repo forgeplus -t v1.0.0 -n "v1.0.0 Stable" -b "Changelog..."
-
-# View a release
-gitlink-cli release +view --owner Gitlink --repo forgeplus -i <version_id>
-```
-
-### CI/CD Operations
-
-```bash
-# List builds
-gitlink-cli ci +list --owner Gitlink --repo forgeplus
-
-# View build log
-gitlink-cli ci +log --owner Gitlink --repo forgeplus -i <build_id>
-
-# Restart a build
-gitlink-cli ci +restart --owner Gitlink --repo forgeplus -i <build_id>
-```
-
-### Search
-
-```bash
-# Search repositories
-gitlink-cli search +repos -k "machine learning"
-
-# Search users
-gitlink-cli search +users -k "zhangsan"
-```
-
-### Raw API
-
-For endpoints not covered by shortcuts, use the Raw API directly:
-
-```bash
-# GET request
-gitlink-cli api GET /users/me
-
-# POST request
-gitlink-cli api POST /Gitlink/forgeplus/issues --body '{"subject":"test","description":"..."}'
-
-# With query parameters
-gitlink-cli api GET /Gitlink/forgeplus/commits --query 'page=1&limit=5'
-```
-
-## Global Parameters
-
-| Parameter | Description | Example |
-|-----------|-------------|---------|
-| `--owner` | Repository owner | `--owner Gitlink` |
-| `--repo` | Repository name | `--repo forgeplus` |
-| `--format` | Output format (json/table/yaml) | `--format json` |
-| `--debug` | Enable debug output | `--debug` |
-
-**Automatic context resolution:** When running inside a git repository, `--owner` and `--repo` are automatically resolved from `git remote origin`.
-
-## Branch Conventions
-
-gitlink-cli supports bidirectional code sync between GitHub and GitLink:
-
-| Platform | Default Branch |
-|----------|---------------|
-| GitHub | `main` |
-| GitLink | `master` |
-
-**Push to GitLink from local:**
-
-```bash
-# Method 1: Use git command directly
-git push gitlink main:master
-
-# Method 2: Configure git remote
-git config remote.gitlink.push refs/heads/main:refs/heads/master
-git push gitlink
-```
-
-## AI Agent Skills
-
-The `skills/` directory contains 12 Agent Skill files for AI-automated GitLink operations.
-
-See [skills/README.md](skills/README.md) for details.
-
-| Skill | Description |
-|-------|-------------|
-| `gitlink-shared` | Authentication, global parameters, safety rules, API notes |
-| `gitlink-repo` | Repository operations (create, view, delete, fork, etc.) |
-| `gitlink-issue` | Issue operations (create, update, close, comment, etc.) |
-| `gitlink-pr` | Pull request operations (create, merge, review, etc.) |
-| `gitlink-branch` | Branch management (create, delete, list, protect, unprotect) |
-| `gitlink-release` | Release management (create, view, delete, etc.) |
-| `gitlink-ci` | CI/CD operations (builds, logs, etc.) |
-| `gitlink-search` | Search (repositories, users, etc.) |
-| `gitlink-org` | Organization management (members, teams, etc.) |
-| `gitlink-user` | User management (profile info, etc.) |
-| `gitlink-pm` | Project management (sprints, kanban, weekly reports, etc.) |
-| `gitlink-workflow` | AI-powered workflows (issue triage, PR review, release notes, etc.) |
-
-## Project Structure
+## 目录结构
 
 ```
 gitlink-cli/
-├── cmd/                      # Cobra command definitions
-│   ├── root.go               # Root command + global flags
-│   ├── auth/                 # Authentication commands
-│   ├── api/                  # Raw API commands
-│   ├── config/               # Configuration commands
-│   └── cmdutil/              # Global utilities
-├── internal/                 # Internal packages
-│   ├── auth/                 # Login, token storage, transport
-│   ├── client/               # HTTP client + pagination
-│   ├── config/               # Config file management
-│   ├── context/              # Git remote resolution
-│   └── output/               # Envelope + formatter
-├── shortcuts/                # Shortcut implementations
-│   ├── common/               # Framework (types, runner)
-│   ├── repo/                 # Repository shortcuts
-│   ├── issue/                # Issue shortcuts
-│   ├── pr/                   # PR shortcuts
-│   ├── branch/               # Branch shortcuts
-│   ├── release/              # Release shortcuts
-│   ├── org/                  # Organization shortcuts
-│   ├── ci/                   # CI shortcuts
-│   ├── search/               # Search shortcuts
-│   ├── user/                 # User shortcuts
-│   └── register.go           # Registration entry point
-├── skills/                   # AI Agent Skills
-│   ├── README.md             # Skills guide
-│   ├── gitlink-shared/       # Shared rules
-│   ├── gitlink-repo/         # Repository skill
-│   ├── gitlink-issue/        # Issue skill
-│   ├── gitlink-pr/           # PR skill
-│   ├── gitlink-pm/           # Project management skill
-│   └── ...
-├── doc/                      # Design documents
-│   ├── Design.md
-│   ├── CODE_SYNC_STRATEGY_FINAL.md
-│   └── ...
-├── main.go
-├── Makefile
-├── go.mod
-└── README.md
+├── README.md                # 本文件
+├── cmd/                     # 入口层（auth/api/config/version）
+├── shortcuts/               # 业务层（19 命令模块 + workflow 工作流引擎）
+├── internal/                # 基础层（client/auth/config/output/context/errors）
+├── skills/                  # 23 个 AI Agent Skill
+├── showcase/                # 在线演示 Dashboard（Go 单二进制）
+├── scripts/                 # 脚本（setup-skills.sh 等）
+├── docs/                    # 设计文档、API 参考
+└── 提交材料/                # ★ 比赛提交材料（变更说明 + 演示录屏/截图）
+    ├── 子赛题一-CLI能力/
+    ├── 子赛题二-Skills/
+    ├── 子赛题三-工作流/
+    └── 子赛题四-科研/
 ```
 
-## Documentation
+---
 
-- [Skills Guide](skills/README.md) — AI Agent Skills detailed documentation
-- [Design Document](doc/design.md) — Architecture design and development plan
+## 技术栈
 
-## FAQ
+- **语言**：Go 1.26（子赛题一）；Markdown/Shell（子赛题二三四）
+- **CLI 框架**：spf13/cobra
+- **密钥存储**：zalando/go-keyring（OS keychain + 文件 fallback）
+- **配置**：gopkg.in/yaml.v3
+- **CI/CD**：GitLink DevOps（建木引擎）自动部署 + GitHub Actions 8 平台发版
 
-### Q: How do I use gitlink-cli in scripts?
+---
 
-Use the `GITLINK_TOKEN` environment variable + `--format json` for structured output:
+## 团队
 
-```bash
-export GITLINK_TOKEN="your-private-token"
-gitlink-cli repo +list --format json | jq '.data.projects[] | .name'
-```
+| 成员 | 负责 |
+|------|------|
+| chroe | CLI 核心模块、Showcase、CI/CD、health/changelog/triage Skill、工作流引擎 |
+| yetja | 批量操作、单元测试、license/repo/org Skill、工作流规则引擎 |
+| caoweiqiong | file/member/watch/star 模块、review Skill、FreeBSD 支持 |
 
-### Q: How does automatic owner/repo resolution work?
+---
 
-When running inside a git repository, the CLI automatically resolves `--owner` and `--repo` from `git remote origin`:
+## 相关链接
 
-```bash
-cd ~/my-gitlink-project
-gitlink-cli issue +list  # Automatically uses the current repository
-```
-
-### Q: What if my token expires?
-
-Re-authenticate:
-
-```bash
-# Username/password login
-gitlink-cli auth login
-
-# Or use a private token (generate at GitLink web → Settings → Private Tokens)
-gitlink-cli auth login --token
-```
-
-### Q: How do I use gitlink-cli in CI/CD or non-interactive environments (e.g. Trae sandbox)?
-
-Set the `GITLINK_TOKEN` environment variable — no `auth login` needed:
-
-```bash
-export GITLINK_TOKEN="your-private-token"
-gitlink-cli repo +list   # Ready to use
-gitlink-cli auth status   # Shows "✓ Logged in via GITLINK_TOKEN environment variable"
-```
-
-Priority: `GITLINK_TOKEN` env var > keyring/file stored token. When the env var is not set, the original interactive login flow works as before.
-
-### Q: What if npm installs successfully but `gitlink-cli` reports a missing binary?
-
-Reinstall first:
-
-```bash
-npm install -g @gitlink-ai/cli
-```
-
-If the error persists, check whether the release page contains the asset for your platform, for example `gitlink-cli_<version>_windows_amd64.zip` on Windows x64. You can also download the binary manually from the release page or build from source with `go install .`.
-
-### Q: Where are credentials stored on Windows?
-
-gitlink-cli uses Windows Credential Manager for secure token storage. If Credential Manager is unavailable, it automatically falls back to file storage (`~/.config/gitlink-cli/credentials`).
-
-### Q: Where can I find the full API reference?
-
-See [skills/gitlink-shared/REFERENCE.md](skills/gitlink-shared/REFERENCE.md).
-
-## License
-
-[MulanPSL-2.0](https://license.coscl.org.cn/MulanPSL2)
+- 上游仓库：<https://gitlink.org.cn/gitlink/gitlink-cli>
+- 本项目（Fork）：<https://gitlink.org.cn/chroe/gitlink-cli>
+- 在线 Showcase：<http://118.31.4.168:9090>
+- Skills 开发指南：[`skills/README.md`](skills/README.md)

README.zh-CN.md

@@ -1,409 +0,0 @@
-# gitlink-cli
-
-[![GitLink](https://img.shields.io/badge/GitLink-Gitlink%2Fgitlink--cli-green)](https://www.gitlink.org.cn/Gitlink/gitlink-cli)
-[![License](https://img.shields.io/badge/License-MulanPSL--2.0-blue.svg)](https://license.coscl.org.cn/MulanPSL2)
-[![Go Version](https://img.shields.io/badge/Go-1.26%2B-blue.svg)](https://golang.org)
-[![npm version](https://img.shields.io/npm/v/@gitlink-ai/cli.svg)](https://www.npmjs.com/package/@gitlink-ai/cli)
-
-[GitLink（确实开源）](https://www.gitlink.org.cn) 官方 CLI 工具 — 为人类和 AI Agent 双重设计。支持 **macOS、Linux、Windows**，覆盖仓库管理、Issue 追踪、Pull Request、CI/CD 和 AI 自动化工作流，包含 40+ 命令和 11 个 AI Agent [Skills](./skills/)。
-
-**[English](./README.md)**
-
-[安装](#安装与快速上手) · [AI Agent Skills](#ai-agent-skills) · [认证](#配置与使用) · [命令](#使用示例) · [贡献](#相关项目)
-
-## 为什么选择 gitlink-cli？
-
-- **Agent-Native 设计** — 开箱即用 11 个结构化 [Skills](./skills/)，兼容 Claude Code — Agent 零配置即可操作 GitLink
-- **广泛覆盖** — 仓库、Issue、PR、分支、Release、CI、组织、搜索、用户 — 核心功能全覆盖
-- **AI 友好 & 优化** — 每条命令都经过真实 Agent 测试，简洁参数、智能默认值、结构化输出
-- **跨平台** — macOS、Linux、Windows (x64/arm64) 全支持，`npm` 一条命令安装
-- **开源零门槛** — 木兰宽松许可证第2版（MulanPSL-2.0），`npm install` 即用
-- **3 分钟上手** — 交互式登录或 `GITLINK_TOKEN` 环境变量，从安装到首次 API 调用仅需 3 步
-- **安全可控** — OS 原生 keychain 凭证存储，`GITLINK_TOKEN` 环境变量支持 CI/CD 和非交互环境，自动 git remote 上下文解析
-- **三层架构** — Shortcuts（人+AI友好）→ Raw API（全覆盖）→ Config（配置管理）
-
-## 功能一览
-
-| 分类 | 能力 |
-|------|------|
-| 📦 仓库 | 列出、创建、Fork、删除仓库，查看仓库信息 |
-| 🐛 Issue | 创建、更新、关闭、批量关闭、评论 Issue |
-| 🔀 PR | 创建、合并、Review Pull Request，查看变更文件 |
-| 🌿 分支 | 创建、删除、保护分支 |
-| 🏷️ 发布 | 创建、查看、删除 Release |
-| 🏢 组织 | 管理组织、成员、团队 |
-| 🔧 CI | 查看构建、日志、CI/CD 操作 |
-| 🔍 搜索 | 搜索仓库、用户 |
-| 👤 用户 | 查看用户资料和信息 |
-| 📋 项目管理 | Sprint 管理、看板、周报 |
-| 🤖 工作流 | AI 驱动的 Issue 分类、PR Review、Release Notes |
-
-## 安装与快速上手
-
-### 前置条件
-
-- Node.js 14+（`npm`/`npx`）— 用于 npm 安装
-- 支持平台：macOS、Linux、Windows（x64/arm64）
-- Go 1.26+ — 仅从源码构建时需要
-
-### 快速上手（人类用户）
-
-> **AI 助手请注意：** 如果你是帮助用户安装的 AI Agent，请直接跳到 [快速上手（AI Agent）](#快速上手ai-agent)，其中包含你需要完成的所有步骤。
-
-#### 安装
-
-选择以下**任一**方式：
-
-**方式 1 — 从 npm 安装（推荐）：**
-
-```bash
-# 安装 CLI
-npm install -g @gitlink-ai/cli
-
-# 安装 CLI Skill（必须，全平台通用）
-gitlink-cli-install-skills
-
-# 也可使用 npx 安装 Skill
-npx skills add ccfos/gitlink-cli/skills -y -g
-```
-
-**方式 2 — 从源码构建：**
-
-需要 Go 1.26+。
-
-```bash
-git clone https://www.gitlink.org.cn/Gitlink/gitlink-cli.git
-cd gitlink-cli
-make install
-
-# 安装 CLI Skill（必须）
-npx skills add ./skills -y -g
-```
-
-> **Windows 用户注意：** 请在 PowerShell 或 CMD 中运行 `npm install -g @gitlink-ai/cli`。从源码构建请使用 `go install .` 代替 `make install`。
-
-#### 配置与使用
-
-```bash
-# 1. 配置（首次使用，交互式引导）
-gitlink-cli config init
-
-# 2. 登录（任选其一）
-gitlink-cli auth login            # 用户名密码（推荐）
-gitlink-cli auth login --token    # 或粘贴私人令牌
-export GITLINK_TOKEN="your-token" # 或设置环境变量（适用于 CI/CD、非交互环境）
-
-# 3. 开始使用
-gitlink-cli repo +list
-```
-
-### 快速上手（AI Agent）
-
-> 以下步骤面向 AI Agent。部分步骤需要用户在浏览器中完成操作。
-
-**第 1 步 — 安装**
-
-```bash
-# 安装 CLI
-npm install -g @gitlink-ai/cli
-
-# 安装 CLI Skill（必须，全平台通用）
-gitlink-cli-install-skills
-```
-
-**第 2 步 — 配置**
-
-```bash
-gitlink-cli config init
-```
-
-**第 3 步 — 登录**
-
-交互环境：
-```bash
-gitlink-cli auth login
-```
-
-非交互环境（CI/CD、Trae 沙箱、MCP 等）：
-```bash
-export GITLINK_TOKEN="your-private-token"
-```
-
-> 获取私人令牌：GitLink 网页端 → 个人设置 → 私人令牌。
-
-**第 4 步 — 验证**
-
-```bash
-gitlink-cli user +me
-```
-
-## 使用示例
-
-### 仓库操作
-
-```bash
-# 列出仓库
-gitlink-cli repo +list
-
-# 查看仓库信息
-gitlink-cli repo +info --owner Gitlink --repo forgeplus
-
-# 创建仓库
-gitlink-cli repo +create -n my-project -d "项目描述"
-
-# Fork 仓库
-gitlink-cli repo +fork --owner Gitlink --repo forgeplus
-```
-
-### Issue 管理
-
-```bash
-# 列出 Issue
-gitlink-cli issue +list --owner Gitlink --repo forgeplus
-
-# 创建 Issue
-gitlink-cli issue +create --owner Gitlink --repo forgeplus -t "Bug: 登录失败" -b "复现步骤..."
-
-# 查看 Issue
-gitlink-cli issue +view --owner Gitlink --repo forgeplus -i 123
-
-# 关闭 Issue
-gitlink-cli issue +close --owner Gitlink --repo forgeplus -i 123
-
-# 预览批量关闭，不修改数据
-gitlink-cli issue +batch-close --owner Gitlink --repo forgeplus --numbers 123,124 --dry-run
-
-# 从 CSV 文件批量关闭 Issue
-gitlink-cli issue +batch-close --owner Gitlink --repo forgeplus --from issues.csv
-
-# 添加评论
-gitlink-cli issue +comment --owner Gitlink --repo forgeplus -i 123 -b "已修复"
-```
-
-### Pull Request
-
-```bash
-# 列出 PR
-gitlink-cli pr +list --owner Gitlink --repo forgeplus
-
-# 创建 PR（同仓库分支）
-gitlink-cli pr +create --owner Gitlink --repo forgeplus -t "feat: 搜索功能" --head feature/search --base master
-
-# 创建 PR（从 Fork 仓库）
-gitlink-cli pr +create --owner Gitlink --repo forgeplus -t "feat: 新功能" --head your_username/forgeplus:feature/my-feature --base master
-
-# 查看 PR
-gitlink-cli pr +view --owner Gitlink --repo forgeplus -i 42
-
-# 合并 PR
-gitlink-cli pr +merge --owner Gitlink --repo forgeplus -i 42
-
-# 查看 PR 变更文件
-gitlink-cli pr +files --owner Gitlink --repo forgeplus -i 42
-```
-
-### 发布管理
-
-```bash
-# 列出 Release
-gitlink-cli release +list --owner Gitlink --repo forgeplus
-
-# 创建 Release
-gitlink-cli release +create --owner Gitlink --repo forgeplus -t v1.0.0 -n "v1.0.0 正式版" -b "更新内容..."
-
-# 查看 Release
-gitlink-cli release +view --owner Gitlink --repo forgeplus -i <version_id>
-```
-
-### 搜索
-
-```bash
-# 搜索仓库
-gitlink-cli search +repos -k "machine learning"
-
-# 搜索用户
-gitlink-cli search +users -k "zhangsan"
-```
-
-### Raw API
-
-Shortcuts 未覆盖的接口可通过 Raw API 直接调用：
-
-```bash
-# GET 请求
-gitlink-cli api GET /users/me
-
-# POST 请求
-gitlink-cli api POST /Gitlink/forgeplus/issues --body '{"subject":"test","description":"..."}'
-
-# 带查询参数
-gitlink-cli api GET /Gitlink/forgeplus/commits --query 'page=1&limit=5'
-```
-
-## 全局参数
-
-| 参数 | 说明 | 示例 |
-|------|------|------|
-| `--owner` | 仓库所有者 | `--owner Gitlink` |
-| `--repo` | 仓库名称 | `--repo forgeplus` |
-| `--format` | 输出格式（json/table/yaml） | `--format json` |
-| `--debug` | 启用调试输出 | `--debug` |
-
-**自动上下文解析**：在 git 仓库目录下，`--owner` 和 `--repo` 会自动从 `git remote origin` 解析。
-
-## 分支约定
-
-gitlink-cli 支持 GitHub 和 GitLink 的代码双向同步：
-
-| 平台 | 主分支 |
-|------|--------|
-| GitHub | `main` |
-| GitLink | `master` |
-
-**本地 push 到 GitLink**：
-
-```bash
-# 方式 1：使用 git 命令
-git push gitlink main:master
-
-# 方式 2：配置 git remote
-git config remote.gitlink.push refs/heads/main:refs/heads/master
-git push gitlink
-```
-
-## AI Agent Skills
-
-`skills/` 目录包含 11 个 Claude Code Agent Skill 文件，支持 AI 自动化操作 GitLink 平台。
-
-详见 [skills/README.md](skills/README.md)
-
-| Skill | 说明 |
-|-------|------|
-| `gitlink-shared` | 认证、全局参数、安全规则、API 注意事项 |
-| `gitlink-repo` | 仓库操作（创建、查看、删除、Fork 等） |
-| `gitlink-issue` | Issue 操作（创建、更新、关闭、评论等） |
-| `gitlink-pr` | Pull Request 操作（创建、合并、Review 等） |
-| `gitlink-release` | 发布管理（创建、查看、删除等） |
-| `gitlink-org` | 组织管理（成员、团队等） |
-| `gitlink-ci` | CI/CD 操作（构建、日志等） |
-| `gitlink-search` | 搜索功能（仓库、用户等） |
-| `gitlink-user` | 用户管理（个人信息等） |
-| `gitlink-pm` | 项目管理（Sprint、看板、周报等） |
-| `gitlink-workflow` | AI 自动化工作流（Issue 分类、PR Review、Release Notes 等） |
-
-## 项目结构
-
-```
-gitlink-cli/
-├── cmd/                      # Cobra 命令定义
-│   ├── root.go               # 根命令 + 全局 flags
-│   ├── auth/                 # 认证命令
-│   ├── api/                  # Raw API 命令
-│   ├── config/               # 配置命令
-│   └── cmdutil/              # 全局工具
-├── internal/                 # 内部包
-│   ├── auth/                 # 登录、Token 存储、Transport
-│   ├── client/               # HTTP 客户端 + 分页
-│   ├── config/               # 配置文件管理
-│   ├── context/              # git remote 解析
-│   └── output/               # Envelope + Formatter
-├── shortcuts/                # Shortcut 实现
-│   ├── common/               # 框架（types, runner）
-│   ├── repo/                 # 仓库 shortcuts
-│   ├── issue/                # Issue shortcuts
-│   ├── pr/                   # PR shortcuts
-│   ├── branch/               # 分支 shortcuts
-│   ├── release/              # Release shortcuts
-│   ├── org/                  # 组织 shortcuts
-│   ├── ci/                   # CI shortcuts
-│   ├── search/               # 搜索 shortcuts
-│   ├── user/                 # 用户 shortcuts
-│   └── register.go           # 注册入口
-├── skills/                   # AI Agent Skills
-│   ├── README.md             # Skills 使用指南
-│   ├── gitlink-shared/       # 共享规则
-│   ├── gitlink-repo/         # 仓库 Skill
-│   ├── gitlink-issue/        # Issue Skill
-│   ├── gitlink-pr/           # PR Skill
-│   ├── gitlink-pm/           # 项目管理 Skill
-│   └── ...
-├── doc/                      # 设计文档
-│   ├── Design.md
-│   ├── CODE_SYNC_STRATEGY_FINAL.md
-│   └── ...
-├── main.go
-├── Makefile
-├── go.mod
-└── README.md
-```
-
-## 文档
-
-- [Skills 使用指南](skills/README.md) — AI Agent Skills 详细说明
-- [设计文档](doc/design.md) — 架构设计和开发计划
-
-## 常见问题
-
-### Q: 如何在脚本中使用 gitlink-cli？
-
-使用 `GITLINK_TOKEN` 环境变量 + `--format json` 获取结构化输出：
-
-```bash
-export GITLINK_TOKEN="your-private-token"
-gitlink-cli repo +list --format json | jq '.data.projects[] | .name'
-```
-
-### Q: 如何自动解析 owner/repo？
-
-在 git 仓库目录下运行命令，CLI 会自动从 `git remote origin` 解析：
-
-```bash
-cd ~/my-gitlink-project
-gitlink-cli issue +list  # 自动使用当前仓库
-```
-
-### Q: Token 过期了怎么办？
-
-重新登录：
-
-```bash
-# 用户名密码登录
-gitlink-cli auth login
-
-# 或使用私人令牌（在 GitLink 网页端 个人设置 → 私人令牌 中生成）
-gitlink-cli auth login --token
-```
-
-### Q: 如何在 CI/CD 或非交互环境（Trae 沙箱等）中使用？
-
-设置 `GITLINK_TOKEN` 环境变量即可，无需 `auth login`：
-
-```bash
-export GITLINK_TOKEN="your-private-token"
-gitlink-cli repo +list   # 直接可用
-gitlink-cli auth status   # 显示 "✓ Logged in via GITLINK_TOKEN environment variable"
-```
-
-Token 优先级：`GITLINK_TOKEN` 环境变量 > keyring/文件存储的 token。不设置环境变量时完全兼容原有交互式登录。
-
-### Q: npm 安装成功但 `gitlink-cli` 提示缺少二进制怎么办？
-
-先尝试重新安装：
-
-```bash
-npm install -g @gitlink-ai/cli
-```
-
-如果仍然失败，请检查 Release 页面是否包含当前平台的资产，例如 Windows x64 对应 `gitlink-cli_<version>_windows_amd64.zip`。也可以从 Release 页面手动下载二进制，或使用 `go install .` 从源码构建。
-
-### Q: Windows 上凭证存储在哪里？
-
-gitlink-cli 使用 Windows Credential Manager 安全存储 Token。如果 Credential Manager 不可用，会自动降级到文件存储（`~/.config/gitlink-cli/credentials`）。
-
-### Q: 如何查看完整的 API 参考？
-
-查看 [skills/gitlink-shared/REFERENCE.md](skills/gitlink-shared/REFERENCE.md)
-
-## 许可证
-
-[MulanPSL-2.0](https://license.coscl.org.cn/MulanPSL2)

README_TASKB.md

@@ -1,335 +0,0 @@
-# gitlink-cli 新增功能使用指南
-
-> 任务B — file / member / watch / star 四大模块，15 个新命令
-
----
-
-## 一、file — 仓库文件操作
-
-### file +list — 列出仓库文件
-
-```bash
-gitlink-cli file +list --owner <owner> --repo <repo> [--ref <分支>] [--search <关键词>]
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--owner` | 否* | 仓库拥有者 |
-| `--repo` | 否* | 仓库名 |
-| `--ref`, `-r` | 否 | 分支/标签/commit SHA |
-| `--search`, `-s` | 否 | 搜索关键词 |
-
-\* 在 git 仓库目录下运行时自动解析，无需手动指定
-
-**示例：**
-
-```bash
-# 列出根目录文件
-gitlink-cli file +list --owner chroe --repo gitlink-cli
-
-# 列出 dev 分支的文件
-gitlink-cli file +list --owner chroe --repo gitlink-cli --ref dev
-
-# 搜索包含 "test" 的文件
-gitlink-cli file +list --owner chroe --repo gitlink-cli --search test
-```
-
-### file +tree — 查看文件树
-
-```bash
-gitlink-cli file +tree --owner <owner> --repo <repo> [--sha <分支>] [--recursive] [--page <n>] [--limit <n>]
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--sha`, `-s` | 否 | 分支/标签/commit，默认 `master` |
-| `--recursive` | 否 | 递归展开所有子目录（`true`/`false`） |
-| `--page`, `-p` | 否 | 页码，默认 `1` |
-| `--limit`, `-l` | 否 | 每页条数，默认 `20` |
-
-**示例：**
-
-```bash
-# 默认 tree 输出（table 格式）
-gitlink-cli file +tree --owner chroe --repo gitlink-cli --sha master
-
-# 递归列出所有文件（JSON 格式）
-gitlink-cli file +tree --owner chroe --repo gitlink-cli --sha master --recursive true --format json
-```
-
-### file +get — 查看文件/目录内容
-
-```bash
-gitlink-cli file +get --owner <owner> --repo <repo> --path <路径> [--ref <分支>]
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--path`, `-p` | **是** | 文件或目录路径 |
-| `--ref`, `-r` | 否 | 分支/标签/commit，默认 `master` |
-
-**示例：**
-
-```bash
-# 查看 README.md 内容
-gitlink-cli file +get --owner chroe --repo gitlink-cli --path README.md
-
-# 查看 src 目录的子条目
-gitlink-cli file +get --owner chroe --repo gitlink-cli --path src
-
-# 查看 dev 分支上的文件
-gitlink-cli file +get --owner chroe --repo gitlink-cli --path main.go --ref dev
-```
-
-### file +create — 新建文件
-
-```bash
-gitlink-cli file +create --owner <owner> --repo <repo> --path <路径> --content <内容> --message <提交信息> [--branch <分支>]
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--path`, `-p` | **是** | 文件路径（含文件名） |
-| `--content`, `-c` | **是** | 文件内容（明文，自动 Base64 编码） |
-| `--message`, `-m` | **是** | Git 提交信息 |
-| `--branch`, `-b` | 否 | 目标分支，默认 `master` |
-
-**示例：**
-
-```bash
-gitlink-cli file +create \
-  --owner caoweiqiong --repo Aether \
-  --path docs/readme.txt \
-  --content "Hello GitLink!" \
-  --message "添加文档" \
-  --branch master
-```
-
-### file +delete — 删除文件
-
-```bash
-gitlink-cli file +delete --owner <owner> --repo <repo> --path <路径> --sha <BlobSHA> --message <提交信息> [--branch <分支>]
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--path`, `-p` | **是** | 要删除的文件路径 |
-| `--sha`, `-s` | **是** | 文件的 blob SHA（从 `file +list` 获取） |
-| `--message`, `-m` | **是** | Git 提交信息 |
-| `--branch`, `-b` | 否 | 目标分支，默认 `master` |
-
-**示例：**
-
-```bash
-# 先列出文件获取 SHA
-gitlink-cli file +list --owner caoweiqiong --repo Aether
-
-# 然后删除指定文件
-gitlink-cli file +delete \
-  --owner caoweiqiong --repo Aether \
-  --path docs/old.txt \
-  --sha abc123def456 \
-  --message "删除过期文档"
-```
-
----
-
-## 二、member — 项目成员管理
-
-### member +list — 列出成员
-
-```bash
-gitlink-cli member +list --owner <owner> --repo <repo> [--keyword <搜索>]
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--keyword`, `-k` | 否 | 按用户名搜索成员 |
-
-**示例：**
-
-```bash
-gitlink-cli member +list --owner chroe --repo gitlink-cli --format table
-
-# 搜索特定成员
-gitlink-cli member +list --owner chroe --repo gitlink-cli --keyword caoweiqiong
-```
-
-输出示例：
-
-```
-id      login         role_name
---      -----         ---------
-149027  chroe         Manager
-141645  caoweiqiong   Developer
-148915  yetja         Developer
-```
-
-### member +add — 添加成员
-
-```bash
-gitlink-cli member +add --owner <owner> --repo <repo> --user-id <数字ID>
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--user-id`, `-u` | **是** | 用户的数字 ID（从 `user +info` 获取） |
-
-**示例：**
-
-```bash
-# 先查到目标用户的 user_id
-gitlink-cli user +info --login zhangsan
-
-# 然后添加
-gitlink-cli member +add --owner caoweiqiong --repo Aether --user-id 123456
-```
-
-### member +remove — 移除成员
-
-```bash
-gitlink-cli member +remove --owner <owner> --repo <repo> --user-id <数字ID>
-```
-
-**示例：**
-
-```bash
-gitlink-cli member +remove --owner caoweiqiong --repo Aether --user-id 123456
-```
-
-### member +update — 修改成员角色
-
-```bash
-gitlink-cli member +update --owner <owner> --repo <repo> --user-id <数字ID> --role <角色>
-```
-
-| 参数 | 必需 | 说明 |
-|------|------|------|
-| `--user-id`, `-u` | **是** | 用户数字 ID |
-| `--role`, `-r` | **是** | 三种角色之一：`Manager` / `Developer` / `Reporter` |
-
-**示例：**
-
-```bash
-gitlink-cli member +update --owner caoweiqiong --repo Aether --user-id 123456 --role Developer
-```
-
----
-
-## 三、watch — 关注仓库
-
-### watch +watch — 关注仓库
-
-```bash
-gitlink-cli watch +watch --owner <owner> --repo <repo>
-```
-
-无需手动传 project-id，命令内部自动解析。
-
-**示例：**
-
-```bash
-gitlink-cli watch +watch --owner chroe --repo gitlink-cli
-# → 返回 {"watched": true}
-```
-
-### watch +unwatch — 取消关注
-
-```bash
-gitlink-cli watch +unwatch --owner <owner> --repo <repo>
-```
-
-**示例：**
-
-```bash
-gitlink-cli watch +unwatch --owner chroe --repo gitlink-cli
-# → 返回 {"watched": false}
-```
-
-### watch +watchers — 查看关注者列表
-
-```bash
-gitlink-cli watch +watchers --owner <owner> --repo <repo>
-```
-
-**示例：**
-
-```bash
-gitlink-cli watch +watchers --owner chroe --repo gitlink-cli --format table
-```
-
----
-
-## 四、star — 点赞仓库
-
-### star +star — 点赞仓库
-
-```bash
-gitlink-cli star +star --owner <owner> --repo <repo>
-```
-
-无需手动传 project-id，命令内部自动解析。
-
-**示例：**
-
-```bash
-gitlink-cli star +star --owner chroe --repo gitlink-cli
-```
-
-### star +unstar — 取消点赞
-
-```bash
-gitlink-cli star +unstar --owner <owner> --repo <repo>
-```
-
-**示例：**
-
-```bash
-gitlink-cli star +unstar --owner chroe --repo gitlink-cli
-```
-
-### star +stars — 查看点赞者列表
-
-```bash
-gitlink-cli star +stars --owner <owner> --repo <repo>
-```
-
-**示例：**
-
-```bash
-gitlink-cli star +stars --owner chroe --repo gitlink-cli --format table
-```
-
----
-
-## 全局参数
-
-所有命令都支持以下全局参数：
-
-| 参数 | 说明 | 示例 |
-|------|------|------|
-| `--owner` | 仓库拥有者（git 目录下自动解析） | `--owner chroe` |
-| `--repo` | 仓库名（git 目录下自动解析） | `--repo gitlink-cli` |
-| `--format` | 输出格式：`json` / `table` / `yaml` | `--format table` |
-| `--debug` | 开启调试输出 | `--debug` |
-
-## 使用技巧
-
-1. **利用自动解析**：在 git 克隆的目录下直接运行，无需写 `--owner` 和 `--repo`：
-
-   ```bash
-   cd my-project
-   gitlink-cli file +list           # 自动识别当前仓库
-   gitlink-cli member +list         # 同上
-   ```
-
-2. **JSON 输出用于脚本**：
-
-   ```bash
-   gitlink-cli file +tree --owner chroe --repo gitlink-cli --sha master --format json | jq '.data.entries[] | .name'
-   ```
-
-3. **file 创建自动 Base64**：传 `--content` 时直接写明文，命令会自动转为 Base64，不用自己编码。
-
-4. **watch/star 透明化**：`watch` 和 `star` 命令会自动查 project-id，你只需关心 owner/repo，与其他命令体验一致。
-

cmd/api/api.go

@@ -4,6 +4,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"net/url"
+	"regexp"
 	"strings"
 
 	"github.com/spf13/cobra"
@@ -13,6 +14,12 @@ import (
 	"github.com/gitlink-org/gitlink-cli/internal/output"
 )
 
+// msysPathRe matches Windows drive paths like "D:/something" that MSYS2
+// auto-converts from Unix paths like "/something". When the api command
+// receives "/v1/owner/repo", Git Bash on Windows may convert it to
+// "C:/Program Files/Git/v1/owner/repo". We detect and revert this.
+var msysPathRe = regexp.MustCompile(`^[A-Za-z]:/`)
+
 func NewAPICmd() *cobra.Command {
 	apiCmd := &cobra.Command{
 		Use:   "api <METHOD> <PATH>",
@@ -36,6 +43,20 @@ func runAPI(c *cobra.Command, args []string) error {
 	method := strings.ToUpper(args[0])
 	path := args[1]
 
+	// Fix MSYS2/Git Bash path auto-conversion on Windows.
+	// "/v1/owner/repo" gets converted to "C:/Program Files/Git/v1/owner/repo".
+	// Detect Windows drive letter prefix and strip it, then restore leading "/".
+	if msysPathRe.MatchString(path) {
+		// Find where the original API path starts (after the git install dir)
+		// by looking for common API prefixes like /v1/, /api/, /users/, etc.
+		for _, prefix := range []string{"/v1/", "/v2/", "/api/", "/users/", "/projects/"} {
+			if idx := strings.Index(path, prefix); idx >= 0 {
+				path = path[idx:]
+				break
+			}
+		}
+	}
+
 	if !strings.HasPrefix(path, "/") {
 		path = "/" + path
 	}

docs/superpowers/plans/2026-06-18-gitlink-review-skill.md

@@ -0,0 +1,767 @@
+# gitlink-review（智能代码审查）Skill 实现计划
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** 新增一个纯 Markdown skill `gitlink-review`，引导 AI agent 对指定 GitLink PR 做多视角 + 对抗式自检的结构化代码审查，并把报告卡作为评论发布。
+
+**Architecture:** 交付物是 `skills/gitlink-review/` 下的 SKILL.md + REFERENCE.md + examples/，无 Go 代码。智能来自 agent 按管道执行（取 diff → 降噪 → 6 视角 → 对抗自检 → 综合 → 预览 → 发布）。发布主路径用已验证可用的 `pr +comment`；行内评论为"reviews API 可用时增强"。用"植入缺陷测试 PR"做端到端真实验证。
+
+**Tech Stack:** Markdown + gitlink-cli（`pr +view` / `pr +files` / `pr +diff` / `pr +comment` / `api POST /pulls/:id/reviews`）。
+
+**Spec:** `docs/superpowers/specs/2026-06-18-intelligent-code-review-design.md`
+
+---
+
+## 前置约束（所有任务适用）
+
+- **`master` 受保护**：所有改动在特性分支 `feat/gitlink-review-skill` 上提交；测试用 fixture 在独立分支 `test/review-fixture` 上（仅供测试，不合并）。
+- **平台**：Windows + Git Bash。路径用正斜杠。
+- **认证**：`gitlink-cli` 已登录（`user +me` 可用）。
+
+---
+
+## Task 1：建立特性分支并纳入 spec/plan
+
+**Files:**
+- Track: `docs/superpowers/specs/2026-06-18-intelligent-code-review-design.md`
+- Track: `docs/superpowers/plans/2026-06-18-gitlink-review-skill.md`
+
+- [ ] **Step 1：从最新 master 建分支**
+
+```bash
+cd "C:/Users/CWQ98/Desktop/演化与运维/gitlink-cli"
+git fetch origin
+git checkout master
+git pull --ff-only origin master
+git checkout -b feat/gitlink-review-skill
+```
+
+预期：位于 `feat/gitlink-review-skill`，与 master 同步。
+
+- [ ] **Step 2：确认 CLI 可用**
+
+```bash
+gitlink-cli version
+gitlink-cli user +me --format json
+```
+
+预期：打印版本；`user +me` 返回 `login: caoweiqiong`。
+
+- [ ] **Step 3：把 spec 和 plan 纳入版本控制**
+
+```bash
+git add docs/superpowers/specs/2026-06-18-intelligent-code-review-design.md \
+        docs/superpowers/plans/2026-06-18-gitlink-review-skill.md
+git commit -m "docs(review): 添加 gitlink-review 设计 spec 与实现计划"
+```
+
+---
+
+## Task 2：创建"植入缺陷"测试 PR（端到端验证载体）
+
+**Files:**
+- Create（在 `test/review-fixture` 分支上，不合并）：`review_fixture.go`
+
+- [ ] **Step 1：从 master 建测试分支**
+
+```bash
+git checkout master
+git checkout -b test/review-fixture
+```
+
+- [ ] **Step 2：写 fixture 文件（含 4 个植入点）**
+
+创建 `review_fixture.go`：
+
+```go
+//go:build ignore
+
+// 本文件为 gitlink-review skill 的测试夹具，故意植入缺陷。DO NOT MERGE。
+package reviewtest
+
+import "os"
+
+var _ = os.Getenv // 仅占位引入，避免 unused 报错（build ignore 下不影响）
+
+// User 占位类型
+type User struct{ Name string }
+
+// 1. 正确性：空指针未判
+func CurrentUser(token string) *User {
+	if token == "" {
+		return nil // token 无效返回 nil
+	}
+	return &User{Name: "cwq"}
+}
+
+func Greet(token string) string {
+	u := CurrentUser(token)
+	return "hello " + u.Name // ← BUG: u 可能为 nil，解引用会 panic
+}
+
+// 2. 安全：硬编码 Token（明显伪造字符串，避免触发真实密钥扫描）
+var AdminToken = "glpat-FAKEFAKEFAKE0000000000"
+
+// 3. 测试：新函数 Welcome，本 PR 无对应 _test.go
+func Welcome(name string) string {
+	if name == "" {
+		return "guest"
+	}
+	return "hi " + name
+}
+
+// 4. 伪阳性：看似除零，但调用方 DoMath 已保证除数非 0
+func Divide(a, b int) int {
+	return a / b
+}
+
+func DoMath(x int) int {
+	if x == 0 {
+		return 0 // 上游保证 x != 0
+	}
+	return Divide(100, x) // 因此 Divide 的除零在真实调用路径上是伪阳性
+}
+```
+
+- [ ] **Step 3：提交并推送**
+
+```bash
+git add review_fixture.go
+git commit -m "test(review): 植入缺陷夹具（正确性/安全/测试/伪阳性）"
+git push origin test/review-fixture
+```
+
+- [ ] **Step 4：创建 PR**
+
+```bash
+gitlink-cli pr +create \
+  --owner chroe --repo gitlink-cli \
+  --head test/review-fixture --base master \
+  --title "test: gitlink-review 植入缺陷夹具（请勿合并）" \
+  --body "gitlink-review skill 端到端验证用 PR。含 4 个植入点：空指针、硬编码 Token、缺测试、伪阳性。验证完成后将关闭不合并。" \
+  --format json
+```
+
+预期：`pull_request_number` 返回一个数（记为 `<PRN>`，如 3）。记录到 `review_fixture_notes.txt`（本地临时）：
+
+```
+TEST_PR_NUMBER=<PRN>
+```
+
+- [ ] **Step 5：验证 diff 可取**
+
+```bash
+gitlink-cli pr +files --owner chroe --repo gitlink-cli --id <PRN> --format json
+gitlink-cli pr +diff   --owner chroe --repo gitlink-cli --id <PRN> --format json | head -c 800
+```
+
+预期：files 含 `review_fixture.go`；diff 含上述代码片段。
+
+---
+
+## Task 3：实测 reviews API（决定行内评论能力）
+
+**Files:** 无（结论写入 Task 5 的 REFERENCE.md）
+
+- [ ] **Step 1：在测试 PR 上探测 POST reviews**
+
+```bash
+gitlink-cli api POST /chroe/gitlink-cli/pulls/<PRN>/reviews \
+  --body '{"body":"gitlink-review probe: 行内评论能力探测","event":"COMMENT"}' 2>&1
+```
+
+- [ ] **Step 2：判断并记录结论**
+
+- 若返回 `{"ok":true,...}`（或含 review id）→ **行内评论可用**，记录实际请求体格式。
+- 若返回 404 / URL 被注入 git exec-path → **不可用（与 triage 记录的 `api POST` bug 一致）**。
+
+把结论写入本地 `review_api_probe.txt`：
+
+```
+REVIEWS_API=<available|unavailable>
+NOTES=<观察到的响应或错误>
+```
+
+- [ ] **Step 3：若可用，再试带行号的行内评论**
+
+```bash
+gitlink-cli api POST /chroe/gitlink-cli/pulls/<PRN>/reviews \
+  --body '{"event":"COMMENT","body":"行内探测","line":22,"path":"review_fixture.go","side":"RIGHT"}' 2>&1
+```
+
+记录是否支持 `line/path/side`（行内定位）。结果并入 `review_api_probe.txt`。
+
+> 此 task 无需 commit（结论是数据，将在 Task 5 固化进 REFERENCE.md）。
+
+---
+
+## Task 4：写 SKILL.md（主管道与方法论）
+
+**Files:**
+- Create: `skills/gitlink-review/SKILL.md`
+
+- [ ] **Step 1：切回特性分支并建目录**
+
+```bash
+git checkout feat/gitlink-review-skill
+```
+
+- [ ] **Step 2：写 SKILL.md**
+
+创建 `skills/gitlink-review/SKILL.md`，**完整内容**如下：
+
+````markdown
+---
+name: gitlink-review
+version: 1.0.0
+description: "智能代码审查：分析 PR diff，多视角评审 + 对抗式自检，输出结构化 Review 意见并作为评论发布。当用户需要审查 GitLink PR、做代码 review、自动生成审查意见时触发。"
+metadata:
+  requires:
+    bins: ["gitlink-cli"]
+  cliHelp: "gitlink-cli pr --help"
+---
+
+# gitlink-review（智能代码审查）
+
+**CRITICAL — 开始前必须先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)，其中包含认证、权限处理和 API 注意事项。**
+**CRITICAL — 所有写入操作（发布评论）前默认先预览、确认后再执行；`--auto` 跳过确认但仍受置信度门控。**
+**CRITICAL — 绝不自动 approve / merge PR。审查只是评论，不做合并决策。**
+**CRITICAL — GitLink 操作只能用 `gitlink-cli`。禁止用 `gh`（GitHub CLI）操作 GitLink 资源。**
+
+> **前置条件：** 先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)；详细检查清单与字段映射见 [`REFERENCE.md`](REFERENCE.md)。
+
+## 概述
+
+本 Skill 引导 AI 对指定 GitLink PR 做结构化、多视角、低误报的代码审查，并把"报告卡"作为评论发布。核心特点：
+
+1. **多视角评审团**：6 个视角并行扫 diff
+2. **对抗式自检**：每条候选发现先自我反驳，成立的才留下（直击 AI 审查误报老大难）
+3. **降噪**：自动跳过生成代码 / vendor / lock / 纯重命名
+4. **可执行**：每条发现带 `file:line` + 修复建议 + 理由
+
+## 命令接口（skill 约定参数，非 gitlink-cli 新增 flag）
+
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--owner/--repo` | 自动从 cwd 解析 | 目标仓库 |
+| `--id` | 必填 | PR 编号（`pull_request_number`，网页 `/pulls/N`） |
+| `--lenses` | 全 6 视角 | 子集，如 `correctness,security` |
+| `--auto` | 关 | 跳过预览确认直接发布（仍受置信度门控） |
+| `--inline` | 关 | 尝试附行内评论（reviews API 可用时） |
+| `--max-findings` | 12 | 报告卡上限 |
+| `--refresh` | 关 | 即使存在旧审查哨兵也重审并更新 |
+
+## 管道（8 步）
+
+### ① 取上下文
+
+```bash
+gitlink-cli pr +view  --owner <owner> --repo <repo> --id <N> --format json   # 元数据
+gitlink-cli pr +files --owner <owner> --repo <repo> --id <N> --format json   # 文件 +/-
+gitlink-cli pr +diff  --owner <owner> --repo <repo> --id <N> --format json   # 核心：diff 内容
+gitlink-cli repo +info --owner <owner> --repo <repo> --format json           # 语言校准
+```
+
+### ② 降噪过滤
+
+按下述"降噪规则"跳过噪音文件；在报告卡"范围"行注明跳过数量，不静默吞掉。
+
+### ③ 多视角分析
+
+对每个启用的视角扫 diff，产出候选发现，每条含：
+
+```json
+{ "lens": "correctness", "severity": "high", "likelihood": "likely",
+  "file": "auth/login.go", "line": 42, "what": "空指针未判",
+  "why": "...", "fix": "...", "confidence": "high" }
+```
+
+### ④ 对抗式自检（质量门）
+
+逐条自问（任一成立即丢弃或降级）：
+
+1. 语境里是否已有保护（外层判空、上游校验、框架机制）？
+2. 是真缺陷还是风格偏好？偏好 → 降级 🔵 nit 或丢弃。
+3. 引用的 API/签名/语言行为是否真实存在？（不确定则不报）
+4. 是否与另一视角重复？
+
+**置信度门控**：`confidence=low` 且 `severity≠high` → 丢弃。
+
+### ⑤ 综合
+
+跨视角去重 → 按 `严重度×likelihood` 排序 → 封顶 `--max-findings`。
+
+### ⑥ 预览（默认）
+
+展示报告卡给用户；确认后发布。`--auto` 跳过。
+
+### ⑦ 发布
+
+```bash
+gitlink-cli pr +comment --owner <owner> --repo <repo> --id <N> \
+  --body "<报告卡全文，含哨兵>"
+```
+
+`--inline` 且 reviews API 可用 → 对 🔴/🟡 发现附行内评论（见 REFERENCE.md 的 API 结论）。
+
+### ⑧ 幂等
+
+发布前检查该 PR 是否已有哨兵 `<!-- gitlink-review v1 ... -->`；有则默认提议"更新"（`--refresh` 才覆盖）。
+
+## 评审团 6 视角
+
+| 视角 | 盯什么 |
+|------|--------|
+| 🔴 正确性 (correctness) | 逻辑错、边界、空值、并发竞态、资源泄漏、错误处理、类型转换 |
+| 🔒 安全 (security) | 注入（SQL/命令/XSS）、鉴权越权、密钥/Token 泄露、路径穿越、不安全反序列化、弱加密 |
+| ⚡ 性能 (performance) | N+1 查询、无谓拷贝、O(n²)/嵌套循环、热路径分配、缺索引 |
+| 🧪 测试 (tests) | 新增/改动代码有无测试、边界用例、断言有效性 |
+| 🧹 可维护性 (maintainability) | 命名、重复、圈复杂度、抽象边界、与既有约定一致 |
+| 🚨 生产风险 (prod-risk) | "合并后凌晨3点哪会炸"——可观测性/日志、回滚、破坏性变更、配置依赖、降级 |
+
+> 各视角的展开检查清单见 [`REFERENCE.md`](REFERENCE.md)。
+
+## 报告卡格式（作为评论发布）
+
+```markdown
+🤖 **gitlink-review 报告**
+
+**结论**：<总体：✅LGTM / ⚠️建议修改 / 🛑有阻塞>（A 🔴阻塞 / B 🟡建议 / C 🔵nit）
+**范围**：<X> 文件，+<A>/-<D>（跳过 <Y> 噪音文件）｜视角：<L>｜自检丢弃：<R>
+
+| 严重度 | 视角 | 位置 | 问题 |
+|:---:|---|---|---|
+| 🔴 | correctness | path/file.go:42 | <一句话> |
+
+### 🔴 阻塞（合并前需处理）
+1. **[correctness] path/file.go:42** — <what>
+   - **为什么**：<why>
+   - **建议**：<fix>
+
+### 🟡 建议
+…
+
+### 🔵 nit
+…
+
+### ✅ 未发现问题的方面
+- <视角>：<结论>
+
+---
+<!-- gitlink-review v1 | pr:<N> | lenses:<L> | refuted:<R> | sha:<head-sha> -->
+*由 gitlink-review skill 生成。*
+```
+
+## 降噪规则（自动跳过）
+
+生成代码（`*.gen.go`/`*.pb.go`/`*_generated.*`/`*.min.js`/dist//build//target/）、第三方（vendor//third_party//node_modules/）、锁文件（go.sum/package-lock.json/yarn.lock/pnpm-lock.yaml/Cargo.lock）、纯重命名/移动、二进制/图片/字体。
+
+## 安全护栏
+
+- 默认预览确认；`--auto` 跳过但仍受置信度门控
+- **绝不自动 approve / merge**
+- 行内评论仅 `--inline` 且 reviews API 可用时发
+- `--max-findings` 封顶防刷屏
+- 评论失败 → 把报告卡原文交用户手动粘贴
+
+## 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| 无 diff | 报"无可审查变更"，不发评论 |
+| PR 不存在/无权限 | 清晰错误，不发评论 |
+| reviews API 404 | 跳过行内，总结评论照发 |
+| diff 过大 | 抽样 + 标注"部分审查（仅 X/Y 文件）" |
+| 评论发布失败 | 输出报告卡原文供手动粘贴 |
+| 全部 low 置信 | 报"未发现高置信问题"，列待人工确认项 |
+
+## 最佳实践
+
+- **先预览再发布**（默认），降低对外噪音
+- **置信度优先**：宁可少报，不要误报
+- **可执行**：每条发现必须带"建议修复"
+- **不越权**：只评论，不合并
+````
+
+- [ ] **Step 3：提交**
+
+```bash
+git add skills/gitlink-review/SKILL.md
+git commit -m "feat(review): 新增 gitlink-review SKILL.md（管道+6视角+对抗自检+报告卡）"
+```
+
+---
+
+## Task 5：写 REFERENCE.md（清单 + 评级 + API 结论）
+
+**Files:**
+- Create: `skills/gitlink-review/REFERENCE.md`
+
+- [ ] **Step 1：读 Task 3 的探测结论**
+
+```bash
+cat review_api_probe.txt
+```
+
+把 `<REVIEWS_API>` 与 `<REVIEWS_NOTES>`（下文占位）替换为真实结论。
+
+- [ ] **Step 2：写 REFERENCE.md**
+
+创建 `skills/gitlink-review/REFERENCE.md`，完整内容（**把 `<REVIEWS_API>` 等占位替换为 Task 3 实测结果**）：
+
+````markdown
+# gitlink-review 参考手册
+
+> 各视角检查清单、严重度评级、降噪规则、reviews API 实测结论、字段映射。
+
+---
+
+## 一、各视角检查清单
+
+### 🔴 正确性 (correctness)
+- 空值/nil 解引用未判
+- 边界：off-by-one、数组越界、空集合
+- 错误未处理或被吞（`err != nil` 缺失、`_ = err`）
+- 并发：数据竞态、缺锁、死锁
+- 资源泄漏：文件/连接/goroutine 未关闭或回收
+- 类型转换/断言未检查
+- 逻辑分支遗漏、return 路径不全
+
+### 🔒 安全 (security)
+- 注入：SQL 拼接、命令、XSS、模板未转义
+- 鉴权/越权：缺权限校验、IDOR
+- 密钥/Token 硬编码或写入日志
+- 路径穿越（`../`、用户输入拼路径）
+- 不安全反序列化、弱加密/弱随机
+- 危险默认值（debug 开关、CORS *）
+
+### ⚡ 性能 (performance)
+- N+1 查询、循环内 IO/查询
+- 无谓拷贝大对象、字符串反复拼接
+- O(n²)/深层嵌套循环
+- 热路径分配、缺缓存
+- 缺索引/全表扫描
+
+### 🧪 测试 (tests)
+- 新增/改动函数有无对应测试
+- 边界与异常用例覆盖
+- 断言是否有效（非 `assert true`）
+- mock 是否合理、是否过度
+
+### 🧹 可维护性 (maintainability)
+- 命名是否达意
+- 重复代码（DRY）
+- 圈复杂度过高、函数过长
+- 抽象边界模糊、职责混杂
+- 与既有代码风格/约定不一致
+
+### 🚨 生产风险 (prod-risk)
+- 破坏性变更（API/DB schema/配置格式）
+- 可观测性：关键路径有无日志/指标
+- 回滚能力：是否可安全回退
+- 配置/环境依赖、启动顺序
+- 降级与限流缺失
+
+---
+
+## 二、严重度 × likelihood 评级
+
+| severity | 含义 | 处理 |
+|----------|------|------|
+| 🔴 high | 阻塞：会导致 bug/安全问题/线上故障 | 合并前需处理 |
+| 🟡 medium | 建议：应修复但不强制阻塞 | 建议处理 |
+| 🔵 low | nit：风格/可读性 | 可选 |
+
+| likelihood | 含义 |
+|------------|------|
+| likely | 真实路径上会发生 |
+| possible | 特定条件下发生 |
+| unlikely | 罕见但可能 |
+
+排序权重：`high×likely` > `high×possible` > `medium×likely` > …。
+
+置信度 `confidence`：high/medium/low。**门控**：`confidence=low && severity≠high → 丢弃`。
+
+---
+
+## 三、降噪规则
+
+跳过：`*.gen.go`、`*.pb.go`、`*_generated.*`、`*.min.js`、`dist/`、`build/`、`target/`、`vendor/`、`third_party/`、`node_modules/`、`go.sum`、`package-lock.json`、`yarn.lock`、`pnpm-lock.yaml`、`Cargo.lock`、纯重命名/移动、二进制/图片/字体。跳过时在报告卡"范围"行计数。
+
+---
+
+## 四、reviews API 实测结论（决定行内评论）
+
+**实测命令**（于测试 PR `<PRN>`）：
+
+```bash
+gitlink-cli api POST /<owner>/<repo>/pulls/<N>/reviews \
+  --body '{"body":"...","event":"COMMENT"}'
+```
+
+**结论**：<REVIEWS_API>（可用 / 不可用）
+
+<REVIEWS_NOTES>（实测观察到的响应或错误）
+
+**设计影响**：
+- 若 **不可用**（如返回 404 / git exec-path 注入 URL）→ 行内评论关闭，仅用 `pr +comment` 发总结评论。`--inline` 被忽略并提示原因。
+- 若 **可用** → `--inline` 时对 🔴/🟡 发现调用上述 API 发行内评论；请求体按实测支持的格式（是否含 `line/path/side`）构造。
+
+---
+
+## 五、字段映射
+
+### pr +view 关键字段
+
+| 字段 | 用途 |
+|------|------|
+| `issue.subject` / `issue.description` | 理解 PR 意图，校准审查重点 |
+| `pull_request.base` / `pull_request.head` | 目标/源分支 |
+| `author.login` | 作者（自审盲区提示） |
+| `files_count` / `commits_count` | 范围概览 |
+
+### pr +diff 关键字段
+
+| 字段 | 用途 |
+|------|------|
+| `files[].name` | 文件路径 |
+| `files[].addition` / `deletion` | 增删行数 |
+| `files[].sha` | 文件 sha（哨兵可选） |
+| diff 正文 | 分析输入 |
+
+**行号映射规则**：报告卡里的 `file:line` 用**文件中的实际行号**（非 diff hunks 的 `+n` 相对行号）。从 diff hunk 头（`@@ -a,b +c,d @@`）推算：实际行号 = `c + (hunk 内相对行)`。
+
+---
+
+## 六、幂等哨兵
+
+```
+<!-- gitlink-review v1 | pr:<N> | lenses:<L> | refuted:<R> | sha:<head-sha> -->
+```
+
+- 发布前用 `pr +view`（或取评论）检查是否已含该哨兵
+- `sha` = 审查所基于的 head commit；PR 有新提交时提示"审查已过期，建议重审"
+- 默认不覆盖；`--refresh` 才重发
+````
+
+- [ ] **Step 3：提交**
+
+```bash
+git add skills/gitlink-review/REFERENCE.md
+git commit -m "docs(review): 新增 REFERENCE.md（视角清单+评级+API实测结论+字段映射）"
+```
+
+---
+
+## Task 6：端到端验证——对测试 PR 跑审查
+
+**Files:** 无（验证 + 捕获输出）
+
+- [ ] **Step 1：按 SKILL.md 对 `<PRN>` 执行审查**
+
+人工/agent 依 SKILL.md 管道执行：取 view/files/diff → 降噪 → 6 视角 → 对抗自检 → 综合 → 生成报告卡。把生成的报告卡全文存到本地 `review_output_<PRN>.md`。
+
+```bash
+gitlink-cli pr +view  --owner chroe --repo gitlink-cli --id <PRN> --format json
+gitlink-cli pr +files --owner chroe --repo gitlink-cli --id <PRN> --format json
+gitlink-cli pr +diff  --owner chroe --repo gitlink-cli --id <PRN> --format json
+```
+
+- [ ] **Step 2：验证 4 个植入点**
+
+断言（必须全部满足，否则回改 SKILL.md/REFERENCE.md 后重跑）：
+
+| # | 植入点 | 期望 | 校验 |
+|---|--------|------|------|
+| 1 | `Greet` 空指针 | 🔴 correctness 命中，含 `review_fixture.go:<行>` 与 `u.Name` | 报告卡中能找到 |
+| 2 | `AdminToken` 硬编码 | 🔒 security 命中 | 报告卡中能找到 |
+| 3 | `Welcome` 缺测试 | 🧪 tests 命中（无 _test.go） | 报告卡中能找到 |
+| 4 | `Divide` 除零（DoMath 已保护） | **被对抗自检丢弃**，`refuted` 计数 ≥1，**不应**出现在发现列表 | 报告卡中无此项 |
+
+- [ ] **Step 3：发布到测试 PR 并验证哨兵**
+
+```bash
+gitlink-cli pr +comment --owner chroe --repo gitlink-cli --id <PRN> \
+  --body "$(cat review_output_<PRN>.md)" --format json
+```
+
+预期：`ok: true`。再 `pr +view` 确认 `comments_count` 增加。
+
+- [ ] **Step 4：若验证不通过，回改并重跑**
+
+任何断言失败 → 修正 SKILL.md 的检查清单/自检规则 → 重新执行 Step 1-3，直到 4 项全过。
+
+---
+
+## Task 7：写 examples/review-workflow.md（真实走查）
+
+**Files:**
+- Create: `skills/gitlink-review/examples/review-workflow.md`
+
+- [ ] **Step 1：基于 Task 6 的真实输出写示例**
+
+创建 `skills/gitlink-review/examples/review-workflow.md`：
+
+````markdown
+# 示例：对植入缺陷 PR 的智能审查（真实数据）
+
+> 基于 `chroe/gitlink-cli` 测试 PR #<PRN>（`test/review-fixture`）于 2026-06-18 实际执行。
+> 该 PR 故意植入 4 个问题用于验证 gitlink-review skill。
+
+---
+
+## Step 1：取上下文
+
+```bash
+gitlink-cli pr +view  --owner chroe --repo gitlink-cli --id <PRN> --format json
+gitlink-cli pr +diff  --owner chroe --repo gitlink-cli --id <PRN> --format json
+```
+
+**范围**：1 文件 `review_fixture.go`，+38/-0，无噪音文件需跳过。
+
+## Step 2：6 视角扫描 + 对抗自检（关键过程）
+
+| 候选发现 | 视角 | 自检结果 |
+|----------|------|----------|
+| `Greet` 中 `u.Name` 解引用 nil | correctness | ✅ 成立（无外层判空）→ 保留 🔴 |
+| `AdminToken` 硬编码 | security | ✅ 成立 → 保留 🔴 |
+| `Welcome` 无对应测试 | tests | ✅ 成立（本 PR 无 _test.go）→ 保留 🟡 |
+| `Divide(a,b)` 除零 | correctness | ❌ **被反驳**：`DoMath` 已保证 `x≠0`，真实调用路径除数非 0 → 丢弃 |
+
+**自检丢弃：1 条（refuted=1）**。
+
+## Step 3：发布报告卡（真实输出）
+
+```bash
+gitlink-cli pr +comment --owner chroe --repo gitlink-cli --id <PRN> \
+  --body "<报告卡全文>"
+```
+
+<此处粘贴 Task 6 生成的真实报告卡全文>
+
+**发布结果**：`ok: true`，PR 评论数 +1。
+
+## Step 4：幂等验证
+
+再次运行（不带 `--refresh`）→ 检测到哨兵 `<!-- gitlink-review v1 | pr:<PRN> | ... -->` → 提示"已存在审查，使用 --refresh 更新"，未重复发布。
+
+## 关键结论
+
+- 3 个真实缺陷被对应视角命中，1 个伪阳性被对抗自检丢弃（refuted=1）
+- 行内评论：<根据 Task 3 结论写"可用已附行内" 或 "API 不可用，仅总结评论">
+````
+
+- [ ] **Step 2：提交**
+
+```bash
+git add skills/gitlink-review/examples/review-workflow.md
+git commit -m "docs(review): 新增真实走查示例 review-workflow.md"
+```
+
+---
+
+## Task 8：更新 README 与 workflow 链接
+
+**Files:**
+- Modify: `skills/README.md`
+- Modify: `skills/gitlink-workflow/SKILL.md`
+
+- [ ] **Step 1：README 智能 skill 表加一行**
+
+在 `skills/README.md` 的"智能 Skills"表追加：
+
+```markdown
+| **gitlink-review** | 智能代码审查 | 分析 PR diff，多视角评审 + 对抗式自检，结构化 Review 意见自动评论 |
+```
+
+并在文档导航/按需处补一行指向 `gitlink-review/SKILL.md`。
+
+- [ ] **Step 2：workflow 的浅层 Code Review 链回本 skill**
+
+在 `skills/gitlink-workflow/SKILL.md` 的"AI 在 PR 流程中的角色 → Code Review"处补一句：
+
+```markdown
+> 深度代码审查请使用 [`../gitlink-review/SKILL.md`](../gitlink-review/SKILL.md)（多视角 + 对抗式自检 + 自动评论）。
+```
+
+- [ ] **Step 3：提交**
+
+```bash
+git add skills/README.md skills/gitlink-workflow/SKILL.md
+git commit -m "docs(review): README 智能表登记 gitlink-review，workflow 链回深度审查"
+```
+
+---
+
+## Task 9：清理测试 PR + 推送特性分支 + 建 PR
+
+**Files:** 无
+
+- [ ] **Step 1：关闭测试 PR（不合并）**
+
+```bash
+gitlink-cli pr +close --owner chroe --repo gitlink-cli --id <PRN>
+```
+
+预期：`ok: true`，PR 状态变 closed（未合并，植入缺陷不进 master）。
+
+- [ ] **Step 2：删本地/远端测试分支**
+
+```bash
+git branch -D test/review-fixture
+git push origin --delete test/review-fixture
+```
+
+- [ ] **Step 3：清理本地临时文件**
+
+```bash
+rm -f review_fixture_notes.txt review_api_probe.txt review_output_*.md
+```
+
+- [ ] **Step 4：推送特性分支**
+
+```bash
+git checkout feat/gitlink-review-skill
+git push origin feat/gitlink-review-skill
+```
+
+- [ ] **Step 5：创建合并到 master 的 PR**
+
+```bash
+gitlink-cli pr +create \
+  --owner chroe --repo gitlink-cli \
+  --head feat/gitlink-review-skill --base master \
+  --title "feat: 新增 gitlink-review 智能代码审查 Skill" \
+  --body "## 目的
+子任务二核心场景：智能代码审查（分析 PR diff，多视角+对抗自检，结构化 Review 自动评论）。
+
+## 交付
+- skills/gitlink-review/：SKILL.md + REFERENCE.md + examples/review-workflow.md
+- skills/README.md、skills/gitlink-workflow/SKILL.md：登记与链接
+- docs/superpowers/：设计 spec + 实现计划
+
+## 验证（真实数据）
+植入缺陷测试 PR 已验证：3 个真实缺陷被对应视角命中，1 个伪阳性被对抗自检丢弃（refuted=1），哨兵幂等，pr +comment 发布成功。reviews API：<填 Task 3 结论>。
+
+## 设计要点
+- 多视角评审团（正确性/安全/性能/测试/可维护/生产风险）
+- 对抗式自检质量门（直击误报）
+- 默认预览确认，绝不自动合并" \
+  --format json
+```
+
+- [ ] **Step 6：记录 PR 号并通知用户**
+
+把返回的 `pull_request_number` 报告给用户，等待其合并（master 受保护）。
+
+---
+
+## 验收（全部满足才算完成）
+
+- [ ] `skills/gitlink-review/` 三件套齐全，结构与 triage/health 一致
+- [ ] Task 6 四个植入点断言全过（3 命中 + 1 丢弃）
+- [ ] REFERENCE.md 的 reviews API 结论为实测结果（非猜测）
+- [ ] 报告卡含哨兵，幂等生效
+- [ ] README 与 workflow 链接已更新
+- [ ] 测试 PR 已关闭未合并，测试分支已删
+- [ ] 特性分支已推送，PR 已创建

docs/superpowers/plans/2026-07-01-gitlink-research-fair.md

@@ -0,0 +1,638 @@
+# gitlink-research-fair Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Build a `gitlink-research-fair` Claude Code Skill that turns any GitLink research repo into a FAIR 体检报告卡 + auto-remediation PR + reproducibility certificate.
+
+**Architecture:** Methodology-type Skill (pure Markdown + gitlink-cli calls, no Go). Report-card-centric: M1 FAIR Checker (5-axis rubric, reuses gitlink-license for license/security) + M2 KG + M3 OpenAlex provenance + M4 Remediation. Validated end-to-end on `songhui18/ICCV2021论文复现`.
+
+**Tech Stack:** Markdown (SKILL.md/REFERENCE.md/examples), gitlink-cli (Go CLI, source build), Mermaid (KG viz), OpenAlex REST (optional enrichment).
+
+**Source of truth:** `docs/superpowers/specs/2026-07-01-gitlink-research-fair-design.md` (read it first).
+
+---
+
+## Environment & Gotchas (engineer must know)
+
+- **Branch:** already on `feat/gitlink-research-fair` (master is protected — never push master). Commit only your own files; leave pre-existing `D README_TASKB.md` / `D gitlink-cli.exe` / `?? dist/` untouched.
+- **Binary:** use the go/bin source build. Verify: `gitlink-cli version` → `gitlink-cli dev`, and `gitlink-cli search --help` shows `+repos`. If commands like `file`/`member` are missing, run `go build -o gitlink-cli.exe .` in the repo root and use `./gitlink-cli.exe`.
+- **Encoding (Windows):** gitlink-cli JSON contains Chinese. When parsing with Python, always set `PYTHONUTF8=1 PYTHONIOENCODING=utf-8`. Never inline Chinese inside `python -c "..."` on this shell (GBK-mangles it) — write a `.py` file instead.
+- **Known GitLink CLI bug:** `file +tree --recursive true` returns empty on some repos. Workaround: `file +tree` top-level → `file +get --path <subdir>` recursively, or use `file +list`.
+- **All examples use `--owner songhui18 --repo ICCV2021论文复现`** unless noted. 处方 (open PR) needs write access → **fork first** (`gitlink-cli repo +fork`); if fork/push fails, fall back to `--no-fork` (write materials to local files).
+- **Verification commands** use `grep` (available in Git Bash). Expected outputs are illustrative; the check is "contains the key string".
+
+---
+
+## File Structure
+
+| File | Responsibility |
+|------|----------------|
+| `skills/gitlink-research-fair/SKILL.md` | Entry doc: frontmatter, CRITICAL headers, command interface, 8-step pipeline, 5-axis rubric, report card template, remediation guardrails, certificate, KG/OpenAlex, error table |
+| `skills/gitlink-research-fair/REFERENCE.md` | Deep reference: FAIR4RS per-principle mapping, scoring algorithm, KG schema (triples), OpenAlex fields, SWH-ID notes, **academic citation library**, honesty caveats (incl. 2 banned claims), live-probe findings |
+| `skills/gitlink-research-fair/examples/research-fair-workflow.md` | Real walkthrough on songhui18: report card + remediation PR + certificate, with contrast run on a licensed repo |
+| `skills/gitlink-research-fair/examples/img/` | Screenshots (GitLink web: opened PR) — optional but recommended for deliverable |
+| `skills/README.md` | Add gitlink-research-fair row (科研辅助类) |
+| `skills/gitlink-workflow/SKILL.md` | Optional cross-link |
+
+---
+
+## Task 1: Scaffold SKILL.md (frontmatter + CRITICAL headers)
+
+**Files:**
+- Create: `skills/gitlink-research-fair/SKILL.md`
+
+- [ ] **Step 1: Create SKILL.md with frontmatter and CRITICAL headers**
+
+Content (exact):
+
+```markdown
+---
+name: gitlink-research-fair
+version: 1.0.0
+description: "科研软件 FAIR 体检：分析 GitLink 科研仓库的可发现/可访问/可互操作/可复用/可复现性，输出体检报告卡，自动开 PR 修复缺口并签发 SWH/commit 锚定的可复现证书。当用户需要评估科研仓库的 FAIR 性与可复现性、生成科研软件体检报告时触发。"
+metadata:
+  requires:
+    bins: ["gitlink-cli"]
+  cliHelp: "gitlink-cli repo --help"
+---
+
+# gitlink-research-fair（科研软件 FAIR 体检）
+
+**CRITICAL — 开始前必须先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)，其中包含认证、权限处理和 API 注意事项。**
+**CRITICAL — 所有写操作（开修复 PR）默认先预览、确认后再执行；`--auto` 跳过预览但仍受护栏约束。**
+**CRITICAL — 绝不自动 merge；绝不 force-push；处方只对 fork 开 PR，绝不碰原仓库。**
+**CRITICAL — GitLink 操作只能用 `gitlink-cli`。禁止用 `gh`（GitHub CLI）操作 GitLink 资源。**
+
+> **前置条件：** 先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)；详细检查清单、评分算法、KG schema 与学术引用见 [`REFERENCE.md`](REFERENCE.md)。
+```
+
+- [ ] **Step 2: Verify frontmatter + CRITICAL lines**
+
+Run: `head -20 skills/gitlink-research-fair/SKILL.md | grep -c -e "name: gitlink-research-fair" -e "description:" -e "CRITICAL"`
+Expected: `6` (1 `name:` + 1 `description:` + 4 `CRITICAL` lines). If `<6`, fix.
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/SKILL.md
+git commit -m "feat(research-fair): scaffold SKILL.md（frontmatter + CRITICAL 头）"
+```
+
+---
+
+## Task 2: SKILL.md — 命令接口 + 8 步管道
+
+**Files:**
+- Modify: `skills/gitlink-research-fair/SKILL.md` (append after the 前置条件 line)
+
+- [ ] **Step 1: Append 概述 + 命令接口 + 8 步管道 sections**
+
+Append (exact content):
+
+```markdown
+## 概述
+
+本 Skill 引导 AI 对任意 GitLink 科研仓库做结构化、有学术标准依据、低误报的"FAIR + 可复现性"诊断，产出**体检报告卡**，并对缺口**自动开 PR 修复**、用 **SWH/commit 锚定**签发**可复现证书**。许可证/安全子项**复用 `gitlink-license`**。四幕剧本（v1）：诊断 → 处方 → 证书。
+
+## 命令接口（skill 约定参数，非 CLI flag）
+
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--owner/--repo` | 自动从 cwd 解析 | 目标科研仓库 |
+| `--lenses` | 全 5 轴 | 子集，如 `F,A,Repro` |
+| `--auto` | 关 | 跳过预览直接开 PR（仍受护栏） |
+| `--no-fork` | 关 | 不开 PR，处方物料落本地 |
+| `--enrich` | 开 | 启用 OpenAlex 溯源（限流自动降级） |
+| `--refresh` | 关 | 即使有旧报告哨兵也重评 |
+
+## 管道（8 步）
+
+### ① 取上下文
+`repo +info`（元数据/license/topics/has_dataset）· `file +list`（关键文件清单）
+
+### ② 深采
+`file +get` 读 LICENSE/CITATION.cff/codemeta.json/README/requirements/Dockerfile · `release +list`（版本）· `commit +list`（SHA）· `issue/pr +list`（协作）· `member +list`（作者）
+
+### ③ FAIR 评分（M1）
+逐项评 5 轴（许可证/安全子项委托 `gitlink-license` 流程）→ `✓/✗/⚠/⊥` + 证据 + 置信度
+
+### ④ KG 构建（M2）
+四实体(Repo/Contributor/File/Commit)+Paper/Dataset/License 节点 → Mermaid 小图 + triples JSON
+
+### ⑤ 溯源富集（M3）
+从 README/CITATION 抽论文 → OpenAlex REST 反查作者/机构（best-effort）
+
+### ⑥ 渲染报告卡（hero）
+等级 + 雷达图 + 5 轴逐项表 + 处方摘要 + 证书栏 + KG 小图 + 溯源栏
+
+### ⑦ 处方（M4）
+对 `✗/⚠` 项生成 CITATION.cff/codemeta.json/LICENSE/复现章节 → 默认预览
+
+### ⑧ 发布
+确认后对 fork 开 PR（PR body 带报告卡摘要）+ 出 SWH/commit 证书；内嵌哨兵
+```
+
+- [ ] **Step 2: Verify the pipeline + interface landed**
+
+Run: `grep -c -e "管道（8 步）" -e "命令接口" -e "### ⑧ 发布" skills/gitlink-research-fair/SKILL.md`
+Expected: `3`
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/SKILL.md
+git commit -m "feat(research-fair): SKILL 命令接口 + 8 步管道"
+```
+
+---
+
+## Task 3: SKILL.md — 5 轴 rubric 表
+
+**Files:**
+- Modify: `skills/gitlink-research-fair/SKILL.md` (append after 管道)
+
+- [ ] **Step 1: Append the 5-axis rubric**
+
+Append (exact):
+
+```markdown
+## 5 轴评分 Rubric
+
+> 对标 **FAIR4RS（Nature Sci Data 2022）+ howfairis 5 维**。诚实声明：这是**适配版**评分（社区尚无认证级自动校验器），不是官方认证。**FAIR ≠ 可复现**，故 Repro 单独成轴。逐条映射与证据见 [`REFERENCE.md`](REFERENCE.md)。
+
+每项判定 `✓满足 / ✗缺失 / ⚠部分 / ⊥GitLink 无该字段(跳过并标注)`，附证据来源 + 置信度。
+
+| 轴 | FAIR4RS 映射 | 检查项（证据） |
+|----|--------------|----------------|
+| **F 可发现** | F1/F2 | F1 清晰描述(repo desc ≥20 字) · F2 话题/关键词(topics) · F3 标准元数据(codemeta.json/.zenodo.json) |
+| **A 可访问** | F1.2/R1.1 | A1 公开(is_public) · A2 有许可证(license_id/LICENSE)[gitlink-license] · A3 有版本发布(release tag) · A4 持久标识(commit SHA/SWH-ID) |
+| **I 可互操作** | R1.1/I | I1 依赖清单(requirements/go.mod/package.json/environment.yml) · I2 许可证机器可读(SPDX/license_id) · I3 元数据标准格式(CITATION.cff/codemeta) |
+| **R 可复用** | R1.2/R2 | R1 溯源/引用(CITATION.cff/README 引用) · R2 README 质量(安装+使用) · R3 源码声明[gitlink-license] · R4 无敏感泄露[gitlink-license] |
+| **Repro 可复现** ⭐差异轴 | （独立于 FAIR） | Rep1 环境锁(锁版本/Dockerfile) · Rep2 数据集说明(has_dataset/README 引用) · Rep3 复现步骤(README 章节) · Rep4 版本固定 · Rep5 入口可执行(main/Makefile/CLI) |
+
+**评分**：轴分 = 满足项 / (总项 − ⊥项)（⚠ 计 0.5）；总评 = 5 轴均分（某轴全 ⊥ 则记 N/A 并排除）；等级 **A🟢≥80 / B🟡60-79 / C🔴<60**。`confidence=low` 标"待人工确认"，不进总评。
+```
+
+- [ ] **Step 2: Verify all 5 axes + scoring present**
+
+Run: `grep -c -e "F 可发现" -e "A 可访问" -e "I 可互操作" -e "R 可复用" -e "Repro 可复现" -e "A🟢≥80" skills/gitlink-research-fair/SKILL.md`
+Expected: `6`
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/SKILL.md
+git commit -m "feat(research-fair): SKILL 5 轴 FAIR rubric 表"
+```
+
+---
+
+## Task 4: SKILL.md — 报告卡模板 + 处方护栏 + 证书 + KG/OpenAlex + 错误降级
+
+**Files:**
+- Modify: `skills/gitlink-research-fair/SKILL.md` (append after rubric)
+
+- [ ] **Step 1: Append report card template**
+
+Append (exact):
+
+```markdown
+## 报告卡格式（hero 产出）
+
+\`\`\`markdown
+🏥 **gitlink-research-fair 科研软件体检报告**
+
+**总体**：<等级> <分>/100（F · A · I · R · Repro）｜版本锚定：commit <sha>（<release 或 无>）
+**范围**：检查 19 项（✓N · ✗N · ⚠N · ⊥N，跳过：<原因>）
+
+### 五轴雷达
+（ASCII 或 Mermaid 雷达：F/A/I/R/Repro）
+
+### 逐项
+| 轴 | 项 | 状态 | 证据 |
+|---|---|:---:|---|
+| <轴> | <id 项名> | ✓/✗/⚠/⊥ | <gitlink-cli 证据> |
+
+### 🔧 处方（可自动修复 N 项）
+- [<轴><id>] <生成物>（从 <来源> 推断）
+
+### 📜 可复现证书
+锚定版本：git+commit <sha> ｜ FAIR: <等级> ｜ Repro: <状态>
+（完整 SWH-ID 需 swh-identify；本次用 commit-SHA 锚定）
+
+### 🔗 溯源（OpenAlex）
+<作者/机构/载体 或 "未找到关联论文，已跳过">
+
+### 🧬 科研关系图（KG）
+\`\`\`mermaid
+<Repo—contributes→Contributor · Repo—licensed?—? · Repo—has→Dataset?>
+\`\`\`
+
+---
+<!-- gitlink-research-fair v1 | repo:<owner>/<repo> | grade:<G> | sha:<head-sha> -->
+*由 gitlink-research-fair skill 生成。*
+\`\`\`
+
+哨兵 `<!-- gitlink-research-fair v1 | repo | grade | sha -->` 用于幂等与"基于哪个 commit"标识。重跑检测旧哨兵 → 默认提议更新（`--refresh` 才覆盖）；`sha` 不匹配 → 提示"报告已过期，建议重评"。
+```
+
+- [ ] **Step 2: Append 处方 + 护栏 + 证书 + KG/OpenAlex + 错误降级**
+
+Append (exact):
+
+```markdown
+## 处方（M4，惊艳闭环）
+
+| 缺口 | 生成物 |
+|------|--------|
+| F3 缺 codemeta.json | 从 repo +info + README 生成 codemeta.json 草稿 |
+| R1 缺 CITATION.cff | 从 README/作者生成 CITATION.cff 草稿 |
+| A2 缺 license | 给 MIT / 木兰 PSL v2 模板（任选） |
+| Rep1 缺依赖锁 | 从 import 扫描建议 requirements.txt |
+| Rep3 缺复现章节 | 从入口/README 生成"复现"章节草稿 |
+
+**安全护栏**：默认全量预览 → 确认 → 对 **fork** 开一个 PR（PR body 带报告卡摘要）；`--auto` 跳过预览但**永不 force-push、永不碰原仓库、永不自动 merge**；`--no-fork` 物料落本地。所有写操作需认证。
+
+## 证书（SWH/commit 锚定）
+
+用 HEAD commit SHA + release tag 锚定版本。若环境有 Python `swh.model` → 算完整 SWH-ID；否则 `git+<commit-SHA>` 锚定，报告卡标注"完整 SWH-ID 需 swh-identify"。证书含：仓库、锚定版本、FAIR 等级、Repro 状态、可引用条目（对标 ACM Artifact Badge）。嵌报告卡证书栏。
+
+## KG + OpenAlex（支撑栏）
+
+- **KG**：四实体(Repo/Contributor/File/Commit)+Paper/Dataset/License；关系 contributes-to/authored/depends-on/cites/licensed-under/version-at。输出 Mermaid 小图 + triples JSON（schema 见 REFERENCE）。
+- **OpenAlex**：README/CITATION 抽论文 → REST `/works` 反查作者/机构/载体。降级：限流或抽不到 → 跳过，标注"未找到关联论文"。
+
+## 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| `file +tree --recursive` 返回空 | 手动递归子目录 |
+| 无 LICENSE | A2 ✗，处方建议模板 |
+| OpenAlex 限流/无果 | 跳过溯源栏标注 |
+| `swh.model` 不可用 | commit-SHA 锚定 + 标注 |
+| 仓库过大 | 抽样文件 + "部分审查"标注 |
+| fork 失败/无写权限 | 处方物料落本地（`--no-fork`） |
+| 二进制是 npm 旧版 | 强制 `./gitlink-cli` 或 `go build` 重建 |
+| 仓库私有/无权限 | 清晰错误，不发报告 |
+| 报告/PR 发布失败 | 报告卡原文交用户手动粘贴 |
+```
+
+- [ ] **Step 3: Verify sentinel + key sections**
+
+Run: `grep -c -e "gitlink-research-fair v1 | repo" -e "## 处方" -e "## 证书" -e "swh.model" skills/gitlink-research-fair/SKILL.md`
+Expected: `4`
+
+- [ ] **Step 4: Commit**
+
+```bash
+git add skills/gitlink-research-fair/SKILL.md
+git commit -m "feat(research-fair): SKILL 报告卡模板+处方护栏+证书+KG/OpenAlex+错误降级"
+```
+
+---
+
+## Task 5: REFERENCE.md — FAIR4RS 映射 + 评分算法 + KG schema + 引用库 + 诚实边界
+
+**Files:**
+- Create: `skills/gitlink-research-fair/REFERENCE.md`
+
+- [ ] **Step 1: Create REFERENCE.md with all deep-reference sections**
+
+Content (exact):
+
+```markdown
+# gitlink-research-fair 参考文档
+
+> 本文件为 SKILL.md 的深度参考。评分细则、KG schema、OpenAlex/SWH 技术细节、学术引用库与诚实边界均在此。
+
+## 一、FAIR4RS 逐条映射
+
+| 原则 | 含义 | 本 Skill 检查项 | GitLink 证据 | 兜底（无字段时） |
+|------|------|----------------|--------------|------------------|
+| F1 | 全局唯一持久标识 | A4 持久标识 | repo identifier/URL + commit SHA | commit-SHA 锚定 |
+| F1.2 | 版本级唯一标识 | A3 有版本发布 | `release +list` | ⊥（无 release 则 ✗） |
+| F2 | 富元数据 | F1 描述/F2 topics/F3 codemeta | `repo +info` desc/topics；`file +get` codemeta.json | README 抽取（标注推断） |
+| R1.1 | 清晰可访问许可证 | A2/I2 | `repo +info` license_id；LICENSE 文件 | gitlink-license 识别 |
+| R1.2 | 详细溯源 | R1 | CITATION.cff/README 引用 | LLM 从 README 推断（标注） |
+| R2 | 对其他软件的限定引用 | R3 | 源码 SPDX/依赖清单 | gitlink-license 维度4 |
+
+## 二、评分算法
+
+- 每项状态：`✓`(1) / `⚠`(0.5) / `✗`(0) / `⊥`(跳过，不计入分母)
+- 轴分 = Σ状态值 / (轴内项数 − ⊥项数) × 100
+- 总评 = 各轴分均值（全 ⊥ 轴记 N/A 排除）
+- 等级：A🟢≥80 / B🟡60-79 / C🔴<60
+- `confidence=low`：标注"待人工确认"，不计入分子分母
+
+## 三、可复现性 checklist（独立于 FAIR）
+
+Rep1 环境锁（requirements 锁版本/Dockerfile/environment.yml）· Rep2 数据集说明（has_dataset/README 引用/下载脚本）· Rep3 复现步骤（README "运行/复现/Quick Start" 章节）· Rep4 版本固定（release tag 或 commit 锚定）· Rep5 入口可执行（main.py/train.py/Makefile/CLI）
+
+## 四、KG schema（triples）
+
+实体：Repo, Contributor, File, Commit, Paper, Dataset, License
+关系（主谓宾三元组）：
+- (Contributor)—contributes-to→(Repo)
+- (Contributor)—authored→(Paper)
+- (Repo)—depends-on→(File/依赖)
+- (Paper)—cites→(Paper)
+- (Repo)—licensed-under→(License)
+- (Repo)—version-at→(Commit)
+- (Repo)—has→(Dataset)
+
+输出：`triples.json`（数组 of {s,p,o}）+ Mermaid `graph LR` 小图。
+
+## 五、OpenAlex 字段
+
+- 端点：`https://api.openalex.org/works?search=<title>` （2025-02 起 freemium，建议带 `mailto` 参数走 polite pool）
+- 取字段：`authorships[].author.display_name`、`authorships[].institutions[].display_name`、`host_venue.display_name`（或 `primary_location.source.display_name`）、`concepts[].display_name`、`doi`
+- 降级：HTTP 429/无结果 → 报告卡溯源栏标"未找到关联论文"
+
+## 六、SWH-ID 说明
+
+- 完整 SWH-ID = `swh:1:dir:<hash>` 或 `swh:1:rev:<hash>`，基于 Merkle DAG，需 `swh.model`（Python）计算或 `swh-identify`
+- 环境无 `swh.model` 时：用 `git+<commit-SHA>` 作版本锚定，报告卡标注"完整 SWH-ID 需 swh-identify"
+- commit SHA 来自 `commit +list`（HEAD）
+
+## 七、学术引用库（"信服"骨架，每条对抗式核验过）
+
+| 支撑点 | 文献 |
+|--------|------|
+| 科研软件 FAIR 原则 F1/F1.2/F2/R1.1/R1.2/R2 可机器校验 | Barker et al. 2022, *Nature Scientific Data*, https://www.nature.com/articles/s41597-022-01710-x |
+| 可复现性危机：2024 ICLR/ICML/NeurIPS 仅 19.5% 提供官方代码 | PaperCoder, arXiv:2504.17192 (Table 9) |
+| 结构化对比表/报告卡范式 | ORKG, Jaradeh et al. K-CAP 2019 |
+| 仓库→RDF 知识图谱 schema（13 实体/47 关系/794 万三元组） | LPWC, ISWC 2023 |
+| SWH-ID 版本锚定（Merkle DAG, git 兼容） | Di Cosmo et al. ICMS 2020, PMC7340894 |
+| 作者/机构溯源用 OpenAlex REST | Priem et al. 2022, arXiv:2205.01833 |
+| 仓库级 KG 问答（四实体+SZZ+Cypher, CoT 50%→90%） | Repo-KG, arXiv:2412.03815 |
+| 科研软件可复现徽章体系 | ACM Artifact Review Badging, https://www.acm.org/publications/policies/artifact-review-badging |
+| 直接竞品 howfairis（5 维，仅 GitHub） | https://github.com/fair-software/howfairis |
+| 战略时机：PWC 不稳定 | TIB 博客 2025-10 "Papers With Code went offline"（单一二手源，pitch 前复核） |
+
+## 八、诚实边界（不可过度宣称）
+
+1. FAIR4RS 自称 **aspirational**；本 Skill 是**自建 checker**（社区尚无认证级校验器），不说"套用现成标准工具"。
+2. **FAIRness ≠ 可复现性**（FAIR 必要非充分）—— Repro 单独成轴。
+3. OpenAlex 自 2025-02 起 freemium（~$1/day、需 key、100 req/s）—— 单仓演示够，批量控量。
+4. **禁止使用两条已证伪论点**：①"MSR 六分类法"、②"FAIR 分高→被引更多"因果。
+5. 文献多跑在 GitHub；"在 GitLink 上复刻"是合理外推，须真机跑通闭环。
+
+## 九、实测结论（live probe，Task 6 回填）
+
+> 此节由 Task 6 在 `songhui18/ICCV2021论文复现` 真机验证后回填：license 状态、README/requirements/releases/commit SHA、各轴实测分、fork→PR 链路是否可用、OpenAlex 是否命中。
+```
+
+- [ ] **Step 2: Verify citations + banned claims + sections**
+
+Run: `grep -c -e "s41597-022-01710-x" -e "2504.17192" -e "禁止使用两条已证伪" -e "## 九、实测结论" skills/gitlink-research-fair/REFERENCE.md`
+Expected: `4`
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/REFERENCE.md
+git commit -m "feat(research-fair): REFERENCE（FAIR4RS映射+评分+KG schema+引用库+诚实边界）"
+```
+
+---
+
+## Task 6: Live probe songhui18 — 回填 REFERENCE 实测结论
+
+**Files:**
+- Modify: `skills/gitlink-research-fair/REFERENCE.md` (§九 实测结论)
+
+- [ ] **Step 1: Probe repo metadata + files**
+
+Run (capture outputs):
+
+```bash
+gitlink-cli repo +info --owner songhui18 --repo ICCV2021论文复现 --format json > /tmp/rf_info.json
+gitlink-cli file +list --owner songhui18 --repo ICCV2021论文复现 --format json > /tmp/rf_files.json
+gitlink-cli release +list --owner songhui18 --repo ICCV2021论文复现 --format json > /tmp/rf_rel.json
+gitlink-cli commit +list --owner songhui18 --repo ICCV2021论文复现 --page 1 --format json > /tmp/rf_commits.json
+```
+
+Expected: each file non-empty, valid JSON with `"ok": true`. If any errors, debug auth/binary first (see Environment & Gotchas).
+
+- [ ] **Step 2: Extract ground truth (license/desc/topics/releases/README/requirements/commit SHA)**
+
+Write `_probe.py` (UTF-8) and run with `PYTHONUTF8=1 PYTHONIOENCODING=utf-8 python _probe.py`:
+
+```python
+# -*- coding: utf-8 -*-
+import json, glob
+def load(p):
+    try:
+        raw=open(p,encoding='utf-8').read(); i=raw.find('{')
+        return json.loads(raw[i:]) if i>=0 else {}
+    except Exception as e:
+        return {'_err':str(e)}
+info=load('/tmp/rf_info.json'); files=load('/tmp/rf_files.json')
+rel=load('/tmp/rf_rel.json'); commits=load('/tmp/rf_commits.json')
+d=info.get('data',info) if isinstance(info,dict) else {}
+print('license_id:', d.get('license_id'))
+print('desc_len:', len(d.get('description') or ''))
+print('topics:', d.get('topics'))
+print('has_dataset:', d.get('has_dataset'))
+print('is_public:', d.get('is_public'))
+# releases
+rd=rel.get('data',rel) if isinstance(rel,dict) else {}
+releases=rd.get('releases') or rd.get('project_releases') or []
+print('releases:', len(releases) if isinstance(releases,list) else releases)
+# files — look for key names
+fd=files.get('data',files) if isinstance(files,dict) else {}
+entries=fd.get('entries') or fd.get('files') or fd.get('trees') or []
+names=[ (e.get('name') or e.get('path') or '') for e in entries ] if isinstance(entries,list) else []
+key=['LICENSE','LICENSE.md','COPYING','README.md','README','requirements.txt','environment.yml','Dockerfile','Makefile','CITATION.cff','codemeta.json','.zenodo.json','go.mod','package.json']
+print('present:', [k for k in key if any(k==n or n.endswith('/'+k) for n in names)])
+print('absent :', [k for k in key if not any(k==n or n.endswith('/'+k) for n in names)])
+# HEAD commit sha
+cd=commits.get('data',commits) if isinstance(commits,dict) else {}
+cl=cd.get('commits') or []
+print('head_sha:', (cl[0].get('sha') if cl and isinstance(cl,list) else None))
+```
+
+Expected: prints real values (adapt field paths if GitLink nests differently — inspect raw JSON). Record: license state (expect absent), README/requirements present?, releases (expect 0), HEAD sha, has_dataset.
+
+- [ ] **Step 3: Backfill REFERENCE §九 with real findings**
+
+Replace the `> 此节由 Task 6...回填` line with real data: license state (expected: absent), README present?, requirements present?, releases count (expected: 0), HEAD commit SHA, has_dataset, and each axis's real ✓/✗ (preview the grade — expect B/C due to no license).
+
+- [ ] **Step 4: Commit**
+
+```bash
+git add skills/gitlink-research-fair/REFERENCE.md
+git commit -m "docs(research-fair): REFERENCE 回填 songhui18 实测结论"
+```
+
+---
+
+## Task 7: Live run — 报告卡 + 处方 PR（fork）+ 证书
+
+**Files:**
+- (produces real artifacts; outputs captured into Task 8's examples doc)
+
+- [ ] **Step 1: Produce the report card following SKILL.md**
+
+Using Task 6 ground truth, walk the 8-step pipeline on `songhui18/ICCV2021论文复现`: score all 5 axes, build KG (Mermaid), try OpenAlex enrichment (search README for paper title/DOI → query api.openalex.org; if none, mark skipped), render the full report card with real values + sentinel + real commit SHA.
+
+Expected: a complete Markdown report card with a real grade (B/C), real axis scores, real evidence per item.
+
+- [ ] **Step 2: Fork + open remediation PR**
+
+```bash
+gitlink-cli repo +fork --owner songhui18 --repo ICCV2021论文复现 --format json
+# clone the fork locally, add generated CITATION.cff + LICENSE + 复现章节, commit, push
+# then:
+gitlink-cli pr +create --owner <你的login> --repo ICCV2021论文复现 \
+  --head <你的login>:feat/fair-remediation --base master \
+  --title "Improve research FAIRness & reproducibility (gitlink-research-fair)" \
+  --body "<报告卡摘要>"
+```
+
+Expected: PR created with a numeric id. **Fallback**: if fork/push fails, use `--no-fork` — write CITATION.cff/LICENSE/复现章节 to a local `_remediation/` dir and note "fork 链路不可用，已降级本地物料" in the examples doc.
+
+- [ ] **Step 3: Verify certificate + capture screenshot**
+
+Confirm the certificate line in the report card has the real HEAD commit SHA (from Task 6) and grade. If `swh.model` importable in python, compute full SWH-ID; else keep `git+<sha>`. Capture a screenshot of the opened PR on GitLink web → save to `skills/gitlink-research-fair/examples/img/pr-songhui18.png` (recommended).
+
+- [ ] **Step 4: Commit (artifacts/screenshots only)**
+
+```bash
+git add skills/gitlink-research-fair/examples/img/  # if screenshot captured
+git commit -m "feat(research-fair): songhui18 实跑报告卡+处方PR+证书（截图）"
+```
+
+(No commit if --no-fork fallback and no screenshot — proceed to Task 8.)
+
+---
+
+## Task 8: examples/research-fair-workflow.md — 真实走查
+
+**Files:**
+- Create: `skills/gitlink-research-fair/examples/research-fair-workflow.md`
+
+- [ ] **Step 1: Write the real walkthrough using Task 6+7 outputs**
+
+Content skeleton (fill with REAL data from Tasks 6-7):
+
+```markdown
+# 示例：科研软件 FAIR 体检（真实数据）
+
+> 基于 `songhui18/ICCV2021论文复现`（论文复现合集，19⭐）于 2026-07-XX 实跑。
+
+## Step 1：取上下文（真实输出摘要）
+- license_id: <实测> ｜ 描述: <长度>字 ｜ topics: <...> ｜ has_dataset: <...>
+- 关键文件：LICENSE=<有/无> · README=<...> · requirements=<...> · CITATION.cff=<...> · releases=<N>
+
+## Step 2：5 轴评分（真实）
+| 轴 | 分 | 关键依据 |
+|---|---|---|
+| F | <实> | ... |
+| A | <实> | 无 license → A2 ✗ |
+| I | <实> | ... |
+| R | <实> | ... |
+| Repro | <实> | ... |
+**总体：<等级> <分>/100**
+
+## Step 3：报告卡（真实全文）
+<贴 Task 7 报告卡全文，含哨兵>
+
+## Step 4：处方 PR
+<PR 链接 / 或 --no-fork 降级说明>，含 CITATION.cff + LICENSE + 复现章节
+
+## Step 5：证书
+锚定 commit <SHA>，FAIR <等级>，Repro <状态>
+
+## 对照：有 license 仓库
+跑 `<对照仓库>`（有 license）→ 预期 A 级，证明评分区分度。
+
+## 关键结论
+- "论文复现"仓库被"可复现性"工具体检：主题共振
+- 无 license = 天然缺口，处方有活干
+- 与 howfairis 差异：GitLink 原生 + 自动修复 + 证书
+```
+
+- [ ] **Step 2: Verify real data (no placeholders)**
+
+Run: `grep -E "TBD|TODO|<等级>|<分>|<实>|<实测>|<SHA>" skills/gitlink-research-fair/examples/research-fair-workflow.md`
+Expected: no matches (all `<...>` placeholders filled with real data). If any remain, fill them from Tasks 6-7.
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/examples/research-fair-workflow.md
+git commit -m "feat(research-fair): examples songhui18 真实走查"
+```
+
+---
+
+## Task 9: 对照仓库跑（证明评分区分度）
+
+**Files:**
+- Modify: `skills/gitlink-research-fair/examples/research-fair-workflow.md` (对照 section)
+
+- [ ] **Step 1: Pick a licensed repo and run the pipeline**
+
+Use `leejt/图神经网络快速开发测试工具` (has license_id=Y) — or `Edgedev/Edge深度学习框架`. Run `repo +info` + `file +list` to confirm it has a LICENSE, then score the 5 axes.
+
+Expected: A2 = ✓ (has license) → A axis notably higher than songhui18; overall grade A (≥80) or clearly higher than songhui18's B/C.
+
+- [ ] **Step 2: Fill the 对照 section with real contrast numbers**
+
+Update the "对照：有 license 仓库" section with: repo name, its grade (expect A), and a one-line contrast ("songhui18 B/C（无 license）vs 对照 A（有 license）→ 评分有区分度").
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/examples/research-fair-workflow.md
+git commit -m "docs(research-fair): 对照仓库证明评分区分度"
+```
+
+---
+
+## Task 10: 注册 + README + workflow 链接 + 最终验收
+
+**Files:**
+- Modify: `skills/README.md`
+- Modify: `skills/gitlink-workflow/SKILL.md` (optional cross-link)
+
+- [ ] **Step 1: Register the skill (so Claude Code's Skill tool discovers it)**
+
+Run: `bash scripts/setup-skills.sh` (creates the junction into `~/.claude/skills/`). Then verify discovery: `ls ~/.claude/skills/ | grep gitlink-research-fair`.
+Expected: `gitlink-research-fair` listed.
+
+- [ ] **Step 2: Add README row**
+
+In `skills/README.md`, add to the 智能 Skills table (near gitlink-review):
+
+```markdown
+| **gitlink-research-fair** | 科研软件 FAIR 体检 | 5 轴 FAIR/可复现评分、报告卡、自动修复 PR、SWH/commit 可复现证书 |
+```
+
+Verify: `grep -c "gitlink-research-fair" skills/README.md` → `≥1`.
+
+- [ ] **Step 3: (Optional) cross-link from gitlink-workflow**
+
+In `skills/gitlink-workflow/SKILL.md`, under a research/repo-health context, add a line linking to `../gitlink-research-fair/SKILL.md`.
+
+- [ ] **Step 4: Final acceptance check (spec §16)**
+
+Verify each (run/read):
+- [ ] `skills/gitlink-research-fair/` 三件套齐全（SKILL.md + REFERENCE.md + examples/）
+- [ ] 5 轴 rubric 每项有证据来源 + 状态判定 + 兜底
+- [ ] songhui18 真跑通：报告卡 + 处方 PR（或 --no-fork 降级）+ 证书，有截图
+- [ ] 对照仓库评分有区分度
+- [ ] 边界（私有/空/无 README）降级正确（至少手动核对 SKILL 错误表覆盖）
+- [ ] REFERENCE 含引用库 + 2 条禁用论点
+- [ ] README 已登记
+
+- [ ] **Step 5: Commit + final state**
+
+```bash
+git add skills/README.md skills/gitlink-workflow/SKILL.md
+git commit -m "docs(research-fair): README 登记 + workflow 链接 + 注册"
+git log --oneline -12
+```
+
+Expected: clean commit history on `feat/gitlink-research-fair`, all artifacts present. Then open a PR to master (per protected-branch workflow).

docs/superpowers/plans/2026-07-01-gitlink-spark.md

@@ -0,0 +1,846 @@
+# gitlink-spark Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Build a `gitlink-spark` Skill + runnable `spark.py` that mines "literature↔code semantic gaps" across arXiv × GitLink × GitHub and outputs an opportunity report with optional fork+issue kickoff.
+
+**Architecture:** `scripts/spark.py` is a standalone data-fusion script (fetches real data → JSON). `SKILL.md` orchestrates an LLM that reads the JSON, semantically matches two gap types (theory-no-impl / demand-no-solution) with evidence triples, renders an opportunity report, and optionally forks+issues to kickoff. Pure functions are unit-tested; the live GNN run is integration validation.
+
+**Tech Stack:** Python 3.11 stdlib only (urllib, subprocess, xml.etree, json, argparse) — no pip deps. gitlink-cli. arXiv Atom API + GitHub Search REST.
+
+**Source of truth:** `docs/superpowers/specs/2026-07-01-gitlink-spark-design.md` (read it first).
+
+---
+
+## Environment & Gotchas (engineer must know)
+
+- **Branch:** on `feat/gitlink-research-fair` (spark is 子任务四第二部分, shares PR #5). Commit only your own files; leave pre-existing `D README_TASKB.md` / `D gitlink-cli.exe` / `?? dist/` / `?? _edge_prescription/` untouched.
+- **Encoding:** all python that touches Chinese gitlink-cli output MUST run with `PYTHONUTF8=1 PYTHONIOENCODING=utf-8`. Never inline Chinese in `python -c` — write a `.py` file.
+- **arXiv must be HTTPS** (`https://export.arxiv.org`); plain HTTP is sandbox-blocked (returns 0 bytes).
+- **GitHub unauthenticated = 10 req/min** → `fetch_github_count` sleeps ~7s between calls. Set `GITHUB_TOKEN` env to raise to 5000/h. Cache by query key.
+- **`gitlink-cli search +issues` returns HTML (broken)** — never use it. Use per-repo `gitlink-cli issue +list --owner X --repo Y --state open` (returns JSON).
+- **`--repo` uses identifier** (ASCII slug), not Chinese display name.
+- **No `pip install`** — spark.py uses stdlib only. Tests run via `python test_spark.py` (assert-based, no pytest).
+
+---
+
+## File Structure
+
+| File | Responsibility |
+|------|----------------|
+| `skills/gitlink-spark/scripts/spark.py` | Standalone data fusion: arXiv + gitlink-cli + GitHub → JSON on stdout. Pure parsers + network fetchers + main() |
+| `skills/gitlink-spark/scripts/test_spark.py` | Assert-based unit tests for pure parsers (parse_arxiv_atom, parse_github_search, extract_method_keywords) |
+| `skills/gitlink-spark/SKILL.md` | 4-stage orchestration, gap taxonomy, GitHub threshold rule, report template, kickoff guardrails, error table |
+| `skills/gitlink-spark/REFERENCE.md` | Gap taxonomy detail, GitHub tiers, LLM prompt template, data-source findings, honesty caveats |
+| `skills/gitlink-spark/examples/spark-图神经网络.md` | Real GNN run: 2-3 gap cards + kickoff screenshot |
+| `skills/README.md` | Add gitlink-spark row |
+| `skills/gitlink-workflow/SKILL.md` | Optional cross-link |
+
+---
+
+## Task 1: Scaffold SKILL.md (frontmatter + CRITICAL + pipeline + command interface)
+
+**Files:**
+- Create: `skills/gitlink-spark/SKILL.md`
+
+- [ ] **Step 1: Create SKILL.md with frontmatter, CRITICAL headers, 概述, 命令接口, 4-stage pipeline**
+
+Content (exact):
+
+```markdown
+---
+name: gitlink-spark
+version: 1.0.0
+description: "文献-代码语义缺口挖掘机：给一个研究领域，跨 arXiv × GitLink × GitHub 三源挖'有理论无实现/有需求无解答'语义缺口，输出空白学术机会报告，可一键 fork+issue 起跑。当用户需要找研究点、发现论文-代码空白、科研选题启发时触发。"
+metadata:
+  requires:
+    bins: ["gitlink-cli"]
+  cliHelp: "python skills/gitlink-spark/scripts/spark.py --help"
+---
+
+# gitlink-spark（文献-代码语义缺口挖掘机）
+
+**CRITICAL — 开始前必须先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)，其中包含认证、权限处理和 API 注意事项。**
+**CRITICAL — 缺口由 LLM 揨断，但每条必须带实证三件套（论文 id / GitLink 查询+命中数 / GitHub total_count）；无实证的缺口必须丢弃。**
+**CRITICAL — 起跑（fork+issue）默认预览确认；绝不自动 merge、绝不 force-push、绝不碰原仓库。**
+**CRITICAL — GitLink 操作只能用 `gitlink-cli`。禁止用 `gh`。**
+
+> **前置条件：** 先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)；缺口分类法、GitHub 阈值、LLM prompt 模板见 [`REFERENCE.md`](REFERENCE.md)。
+
+## 概述
+
+给一个研究领域，跨 **arXiv（学术）× GitLink（中文生态）× GitHub（全球）** 三源挖两类语义缺口，输出**空白学术机会报告**。`scripts/spark.py` 抓真实数据（JSON），LLM 做语义匹配并附实证三件套。与 `gitlink-research-fair`（评估已有）组成"科研辅助双联装"——本 skill 负责**发现空白**。
+
+## 命令接口
+
+数据融合脚本（可独立运行）：
+
+```bash
+python skills/gitlink-spark/scripts/spark.py --field "图神经网络" [--max-papers 10] [--gap-type both|theory|demand] [--github-token $GITHUB_TOKEN]
+# → stdout: 融合 JSON {papers, gitlink_repos, gitlink_issues, github_counts}
+```
+
+skill 约定参数（非 CLI flag）：
+
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--field` | 必填 | 研究领域 |
+| `--gap-type` | `both` | `theory` / `demand` / `both` |
+| `--max-papers` | 10 | arXiv 抓取上限（控 GitHub 调用） |
+| `--auto` | 关 | 跳过预览直接起跑（仍受护栏） |
+| `--no-fork` | 关 | 只出报告，不起跑 |
+
+## 管道（4 阶段）
+
+### ① 学术采
+`spark.py` 调 arXiv HTTPS API 抓领域近 90 天论文（标题/摘要/arxiv id/方法关键词）
+
+### ② GitLink 采
+`spark.py` 调 `gitlink-cli search +repos` 抓领域仓库；对每个仓库 `issue +list --state open` 抓 open issue（**不用 search +issues**，它返回 HTML）
+
+### ③ 全球对照
+`spark.py` 调 GitHub Search API 对每个论文方法查 `total_count` + Top3 仓库（限流+缓存）
+
+### ④ 缺口匹配（LLM）+ 报告 + 起跑
+读 spark.py 的 JSON → 语义匹配两类缺口（每张带实证三件套）→ 渲染机会报告 → 可选 fork+issue 起跑
+```
+
+- [ ] **Step 2: Verify**
+
+Run: `head -30 skills/gitlink-spark/SKILL.md | grep -c -e "name: gitlink-spark" -e "description:" -e "CRITICAL" -e "管道（4 阶段）"`
+Expected: `6`
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-spark/SKILL.md
+git commit -m "feat(spark): scaffold SKILL.md（frontmatter+CRITICAL+4阶段管道）"
+```
+
+---
+
+## Task 2: SKILL.md — gap taxonomy + evidence triples + GitHub thresholds + report + guardrails
+
+**Files:**
+- Modify: `skills/gitlink-spark/SKILL.md` (append after 管道)
+
+- [ ] **Step 1: Append gap taxonomy, thresholds, report template, kickoff, error table**
+
+Append (exact):
+
+```markdown
+## 两类缺口 + 实证三件套（信服核心）
+
+每张缺口卡**必须**带齐三件套，否则丢弃（防 LLM 编造）：
+
+### 类型 A：有理论无实现（paper → code gap）
+- **三件套**：① 论文 arxiv id + 标题 + 发表日期 ② GitLink 搜索查询串 + 命中数（0/极少） ③ GitHub total_count + Top 仓库
+- LLM 判定：论文提出方法 M；GitLink 实现 0/极少；GitHub 按下方阈值分级
+
+### 类型 B：有需求无解答（open issue → applied gap）
+- **三件套**：① issue URL + 主题 + 讨论人数/状态 ② GitLink 无现成实现解此痛点 ③ GitHub 是否有成熟开源解
+- 降噪：LLM 只挑"研究性痛点"（性能/可扩展/新场景），排除"安装报错"等使用问题
+
+## GitHub 全球对照阈值（诚实核心，硬需求）
+
+防止"GitLink 0 ≠ 全球空白"误导。对每个"理论无实现"候选按 GitHub total_count 分级：
+
+| GitHub total_count | 分级 | 报告行为 |
+|--------------------|------|----------|
+| `< 10` | 全球稀缺（真空白） | 报为高价值缺口 |
+| `10–50` | 新兴（部分空白） | 报为中等缺口（"GitLink 空白，全球新兴"） |
+| `≥ 50` | 全球已成熟 | **不报为空白**，列入"✅ 已诚实排除" |
+
+宁可少报，不误报机会。
+
+## 机会报告格式（hero）
+
+````markdown
+⚡ **gitlink-spark 机会报告：<field>**
+
+学术采：arXiv 近 90 天 N 篇 ｜ GitLink 仓库 M 个 ｜ GitHub 全球基线已对照
+生成时间：YYYY-MM-DD
+
+### 🧩 缺口 1 · 有理论无实现 [全球稀缺·高价值]
+**论文**：[arxiv:<id>] "<title>" (<date>)
+**方法关键词**：<...>
+**GitLink**：search "<query>" → **0 命中**（查询串留底）
+**GitHub 全球**：total_count = **N**（Top: <repo> <stars>⭐）→ 稀缺
+**机会建议**：<LLM 一句话>
+**起跑**：[按钮] fork 基准 <repo> → 创建 issue 粘论文伪代码
+
+### 🧩 缺口 2 · 有需求无解答 [应用机会]
+**Issue**：<repo>#<n> "<subject>"（N 人讨论, open）
+**痛点**：<LLM 归纳>
+**GitLink / GitHub**：均无成熟解
+**机会建议**：<LLM 一句话>
+
+### ✅ 已诚实排除（非空白）
+- 论文 Y：GitLink 虽 0，但 GitHub 已 N 个 → 全球已成熟，不报
+
+---
+<!-- gitlink-spark v1 | field:<field> | gaps:<N> | date:<YYYY-MM-DD> -->
+*由 gitlink-spark skill 生成。*
+````
+
+## 起跑动作 + 护栏
+
+选定一张"理论无实现"缺口卡 → 确认 →
+1. `gitlink-cli repo +fork` 最近基准（GitHub Top 仓库或 GitLink 最近实现）
+2. LLM 从 arXiv 论文抓 Algorithm/Pseudocode 节
+3. `gitlink-cli issue +create` 在 fork 建复现 todo issue（body 粘伪代码 + 报告卡摘要）
+
+**护栏**：默认预览；`--auto` 跳过但**永不 force-push、永不碰原仓库、永不自动 merge**；`--no-fork` 只出报告。
+
+## 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| arXiv 空/超时 | HTTPS 重试；仍空降级用既有论文 |
+| `search +issues` 返回 HTML | 不用，改逐仓库 `issue +list` |
+| GitHub 未认证限流(10/min) | spark.py sleep ~7s；建议设 `GITHUB_TOKEN` |
+| GitHub 查询失败 | 该论文标"对照失败"，不进缺口判定 |
+| OpenAlex 503 | 跳过引用富集 |
+| LLM 缺口无三件套 | 置信度门控丢弃 |
+| fork/issue 起跑失败 | 输出 fork 目标 + 伪代码文本供手动起跑 |
+```
+
+- [ ] **Step 2: Verify**
+
+Run: `grep -c -e "三件套" -e "GitHub total_count" -e "已诚实排除" -e "起跑动作" skills/gitlink-spark/SKILL.md`
+Expected: `4`
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-spark/SKILL.md
+git commit -m "feat(spark): SKILL 缺口分类法+GitHub阈值+报告模板+起跑护栏+降级"
+```
+
+---
+
+## Task 3: spark.py — parse_arxiv_atom + test (TDD)
+
+**Files:**
+- Create: `skills/gitlink-spark/scripts/spark.py`
+- Create: `skills/gitlink-spark/scripts/test_spark.py`
+
+- [ ] **Step 1: Write the failing test**
+
+`test_spark.py` (exact):
+
+```python
+# -*- coding: utf-8 -*-
+"""Assert-based unit tests for spark.py pure parsers. Run: python test_spark.py"""
+import sys, os
+sys.path.insert(0, os.path.dirname(__file__))
+from spark import parse_arxiv_atom, parse_github_search, extract_method_keywords
+
+SAMPLE_ARXIV = """<?xml version="1.0" encoding="UTF-8"?>
+<feed xmlns="http://www.w3.org/2005/Atom">
+  <entry>
+    <id>http://arxiv.org/abs/2403.12345v1</id>
+    <title>Graph Attention Networks with Sparse Transformers</title>
+    <summary>We propose a new graph attention mechanism using sparse attention.</summary>
+    <published>2024-03-15T00:00:00Z</published>
+  </entry>
+  <entry>
+    <id>http://arxiv.org/abs/2404.99999v2</id>
+    <title>Federated Learning on Heterogeneous Graphs</title>
+    <summary>A federated approach for heterogeneous graph neural networks.</summary>
+    <published>2024-04-20T00:00:00Z</published>
+  </entry>
+</feed>"""
+
+def test_parse_arxiv_atom():
+    papers = parse_arxiv_atom(SAMPLE_ARXIV)
+    assert len(papers) == 2, f"expected 2 papers, got {len(papers)}"
+    assert papers[0]["arxiv_id"] == "2403.12345v1", papers[0]["arxiv_id"]
+    assert "Graph Attention" in papers[0]["title"]
+    assert papers[0]["published"] == "2024-03-15"
+    assert "sparse" in papers[0]["abstract"].lower()
+    print("test_parse_arxiv_atom OK")
+
+def test_parse_github_search():
+    import json as _j
+    sample = _j.dumps({"total_count": 1543, "items": [{"full_name": "a/b", "stargazers_count": 3534}]})
+    res = parse_github_search(sample)
+    assert res["total_count"] == 1543
+    assert res["top"][0]["full_name"] == "a/b"
+    assert res["top"][0]["stars"] == 3534
+    print("test_parse_github_search OK")
+
+def test_extract_method_keywords():
+    kws = extract_method_keywords("Graph Attention Networks", "We propose a sparse attention mechanism for graphs.", max_k=5)
+    assert "graph" in kws and "attention" in kws
+    assert "propose" not in kws  # 'propose' is in the stop set, filtered out
+    print("test_extract_method_keywords OK")
+
+if __name__ == "__main__":
+    test_parse_arxiv_atom()
+    test_parse_github_search()
+    test_extract_method_keywords()
+    print("ALL TESTS PASSED")
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run: `cd skills/gitlink-spark/scripts && python test_spark.py`
+Expected: FAIL with `ModuleNotFoundError: No module named 'spark'`
+
+- [ ] **Step 3: Write minimal spark.py with parse_arxiv_atom (+ stubs for the other two so import works)**
+
+`spark.py` (exact):
+
+```python
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""gitlink-spark data fusion: arXiv x GitLink x GitHub -> JSON on stdout. Stdlib only."""
+import argparse, json, os, sys, time, subprocess, urllib.request, urllib.parse, re
+from xml.etree import ElementTree as ET
+
+ARXIV_ENDPOINT = "https://export.arxiv.org/api/query"
+GITHUB_ENDPOINT = "https://api.github.com/search/repositories"
+
+_NS = {"a": "http://www.w3.org/2005/Atom"}
+
+def parse_arxiv_atom(xml_text):
+    """Parse arXiv Atom feed -> list of {arxiv_id, title, abstract, published}."""
+    root = ET.fromstring(xml_text)
+    papers = []
+    for e in root.findall("a:entry", _NS):
+        aid = (e.find("a:id", _NS).text or "").strip().split("/")[-1]
+        title = re.sub(r"\s+", " ", (e.find("a:title", _NS).text or "").strip())
+        summary = re.sub(r"\s+", " ", (e.find("a:summary", _NS).text or "").strip())
+        pub = (e.find("a:published", _NS).text or "")[:10]
+        papers.append({"arxiv_id": aid, "title": title, "abstract": summary, "published": pub})
+    return papers
+
+def parse_github_search(json_text):
+    return {"total_count": 0, "top": []}  # stub — implemented in Task 4
+
+def extract_method_keywords(title, abstract, max_k=5):
+    return []  # stub — implemented in Task 4
+
+def main():
+    pass  # implemented in Task 7
+
+if __name__ == "__main__":
+    main()
+```
+
+- [ ] **Step 4: Run test — arxiv test passes, others fail on stubs**
+
+Run: `cd skills/gitlink-spark/scripts && python test_spark.py`
+Expected: `test_parse_arxiv_atom OK`, then FAIL on `test_parse_github_search` (top[0] index error on empty). This confirms arxiv parser works; stubs next.
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add skills/gitlink-spark/scripts/spark.py skills/gitlink-spark/scripts/test_spark.py
+git commit -m "feat(spark): spark.py parse_arxiv_atom + test（TDD）"
+```
+
+---
+
+## Task 4: spark.py — parse_github_search + extract_method_keywords (real impl)
+
+**Files:**
+- Modify: `skills/gitlink-spark/scripts/spark.py` (replace the two stubs)
+
+- [ ] **Step 1: Replace the two stub functions with real implementations**
+
+Replace `def parse_github_search(json_text): ...` and `def extract_method_keywords(...): ...` with:
+
+```python
+def parse_github_search(json_text):
+    """Parse GitHub search JSON -> {total_count, top:[{full_name, stars}]}."""
+    d = json.loads(json_text)
+    return {
+        "total_count": d.get("total_count", 0),
+        "top": [{"full_name": r.get("full_name"), "stars": r.get("stargazers_count")}
+                for r in (d.get("items") or [])[:3]],
+    }
+
+def extract_method_keywords(title, abstract, max_k=5):
+    """Crude keyword extraction for GitHub/arXiv query."""
+    text = (title + " " + abstract).lower()
+    stop = {"the", "a", "an", "of", "for", "and", "to", "in", "on", "with", "via",
+            "based", "using", "by", "from", "as", "is", "are", "we", "our", "this",
+            "that", "propose", "proposed", "paper", "method", "approach", "novel", "new"}
+    tokens = re.findall(r"[a-z][a-z0-9-]+", text)
+    seen = set(); out = []
+    for t in tokens:
+        if t in stop or len(t) < 3 or t in seen:
+            continue
+        seen.add(t); out.append(t)
+        if len(out) >= max_k:
+            break
+    return out
+```
+
+- [ ] **Step 2: Run all tests — expect ALL PASS**
+
+Run: `cd skills/gitlink-spark/scripts && python test_spark.py`
+Expected: `ALL TESTS PASSED` (all 3 tests)
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-spark/scripts/spark.py
+git commit -m "feat(spark): parse_github_search + extract_method_keywords 实现（3 测试全过）"
+```
+
+---
+
+## Task 5: spark.py — fetch_arxiv (network)
+
+**Files:**
+- Modify: `skills/gitlink-spark/scripts/spark.py` (add fetch_arxiv before main)
+
+- [ ] **Step 1: Add fetch_arxiv**
+
+Insert before `def main()`:
+
+```python
+def fetch_arxiv(field, max_papers=10):
+    """Search arXiv (HTTPS) for recent papers in field. Returns list of paper dicts."""
+    q = urllib.parse.quote(f'abs:"{field}"')
+    url = (f"{ARXIV_ENDPOINT}?search_query={q}&max_results={max_papers}"
+           f"&sortBy=submittedDate&sortOrder=descending")
+    with urllib.request.urlopen(url, timeout=30) as r:
+        papers = parse_arxiv_atom(r.read().decode("utf-8", "replace"))
+    for p in papers:
+        p["method_keywords"] = extract_method_keywords(p["title"], p["abstract"])
+    return papers
+```
+
+- [ ] **Step 2: Smoke test fetch_arxiv (live network)**
+
+Run:
+```bash
+cd skills/gitlink-spark/scripts && PYTHONUTF8=1 python -c "
+from spark import fetch_arxiv
+ps = fetch_arxiv('graph neural network', max_papers=2)
+assert len(ps) >= 1, 'no papers'
+p = ps[0]
+assert p['arxiv_id'] and p['title'] and p['published']
+assert isinstance(p['method_keywords'], list)
+print('OK', p['arxiv_id'], '|', p['title'][:50])
+"
+```
+Expected: `OK 2504.xxxxx | <recent GNN paper title>` (a real recent arxiv id). If 0 bytes, confirm HTTPS (not HTTP).
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-spark/scripts/spark.py
+git commit -m "feat(spark): fetch_arxiv（arXiv HTTPS 网络层）"
+```
+
+---
+
+## Task 6: spark.py — fetch_gitlink_repos + fetch_gitlink_issues
+
+**Files:**
+- Modify: `skills/gitlink-spark/scripts/spark.py`
+
+- [ ] **Step 1: Add _gitlink helper + the two fetchers**
+
+Insert before `def main()`:
+
+```python
+def _gitlink(*args):
+    """Run gitlink-cli with json output; return parsed dict (UTF-8 safe)."""
+    r = subprocess.run(["gitlink-cli"] + list(args) + ["--format", "json"],
+                       capture_output=True, text=True, encoding="utf-8",
+                       errors="replace", timeout=60)
+    raw = r.stdout
+    i = raw.find("{")
+    return json.loads(raw[i:]) if i >= 0 else {}
+
+def fetch_gitlink_repos(field):
+    """gitlink-cli search +repos -> list of {owner, repo(identifier), name, desc, topics}."""
+    d = _gitlink("search", "+repos", "-k", field)
+    projs = d.get("data", {}).get("projects", []) or []
+    out = []
+    for p in projs:
+        out.append({
+            "owner": (p.get("author") or {}).get("login"),
+            "repo": p.get("identifier"),
+            "name": p.get("name"),
+            "desc": p.get("description"),
+            "topics": [t.get("name") if isinstance(t, dict) else t for t in (p.get("topics") or [])],
+        })
+    return out
+
+def fetch_gitlink_issues(repos, max_per_repo=10):
+    """Per-repo issue +list (open) -> list of {repo, number, subject, status, participants}.
+    Works around search +issues returning HTML."""
+    out = []
+    for r in repos:
+        if not (r.get("owner") and r.get("repo")):
+            continue
+        d = _gitlink("issue", "+list", "--owner", r["owner"], "--repo", r["repo"], "--state", "open")
+        data = d.get("data", {}) or {}
+        issues = data.get("issues") or []
+        for it in issues[:max_per_repo]:
+            st = (it.get("status") or {})
+            if st.get("name") == "关闭":
+                continue
+            out.append({
+                "repo": f'{r["owner"]}/{r["repo"]}',
+                "number": it.get("project_issues_index") or it.get("number"),
+                "subject": it.get("subject"),
+                "status": st.get("name"),
+                "participants": it.get("participants_count") or 0,
+            })
+    return out
+```
+
+- [ ] **Step 2: Smoke test (live)**
+
+Run:
+```bash
+cd skills/gitlink-spark/scripts && PYTHONUTF8=1 PYTHONIOENCODING=utf-8 python -c "
+from spark import fetch_gitlink_repos, fetch_gitlink_issues
+repos = fetch_gitlink_repos('图神经网络')
+print('repos:', len(repos))
+if repos: print('  sample:', repos[0]['owner'], '/', repos[0]['repo'])
+iss = fetch_gitlink_issues(repos[:2])
+print('issues(from first 2 repos):', len(iss))
+"
+```
+Expected: `repos: N` (N≥1, includes GraphGallery-class), `issues: M`. If `repos: 0`, the field keyword missed — retry with `'graph neural'`. If issues JSON parse fails, inspect gitlink-cli `issue +list` structure and adapt field names.
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-spark/scripts/spark.py
+git commit -m "feat(spark): fetch_gitlink_repos + fetch_gitlink_issues（逐仓库，绕开 search+issues HTML）"
+```
+
+---
+
+## Task 7: spark.py — fetch_github_count + main() + end-to-end smoke
+
+**Files:**
+- Modify: `skills/gitlink-spark/scripts/spark.py`
+
+- [ ] **Step 1: Add fetch_github_count (cache+throttle) and implement main()**
+
+Replace `def main(): pass` and add fetch_github_count before it:
+
+```python
+_GH_CACHE = {}
+
+def fetch_github_count(query, token=None, throttle=True):
+    """GitHub search total_count + top3 for a query. Caches + throttles (10/min unauth)."""
+    if query in _GH_CACHE:
+        return _GH_CACHE[query]
+    url = f"{GITHUB_ENDPOINT}?q={urllib.parse.quote(query)}&per_page=3&sort=stars"
+    req = urllib.request.Request(url, headers={"Accept": "application/vnd.github+json",
+                                               "User-Agent": "gitlink-spark/1.0"})
+    if token:
+        req.add_header("Authorization", f"Bearer {token}")
+    try:
+        with urllib.request.urlopen(req, timeout=25) as r:
+            res = parse_github_search(r.read().decode("utf-8", "replace"))
+    except Exception as e:
+        res = {"total_count": None, "top": [], "error": str(e)[:80]}
+    if throttle and not token:
+        time.sleep(7)  # unauthenticated = 10 req/min
+    _GH_CACHE[query] = res
+    return res
+
+def main():
+    ap = argparse.ArgumentParser(description="gitlink-spark data fusion")
+    ap.add_argument("--field", required=True)
+    ap.add_argument("--max-papers", type=int, default=10)
+    ap.add_argument("--gap-type", default="both", choices=["both", "theory", "demand"])
+    ap.add_argument("--github-token", default=os.environ.get("GITHUB_TOKEN"))
+    args = ap.parse_args()
+
+    papers = fetch_arxiv(args.field, args.max_papers)
+    grepos = fetch_gitlink_repos(args.field)
+    gissues = fetch_gitlink_issues(grepos) if args.gap_type in ("both", "demand") else []
+    gh_counts = {}
+    if args.gap_type in ("both", "theory"):
+        for p in papers:
+            q = p["title"][:60]  # primary query = paper title (truncated)
+            gh_counts[q] = fetch_github_count(q, args.github_token)
+
+    out = {
+        "field": args.field,
+        "papers": papers,
+        "gitlink_repos": grepos,
+        "gitlink_issues": gissues,
+        "github_counts": gh_counts,
+    }
+    json.dump(out, sys.stdout, ensure_ascii=False, indent=2)
+    sys.stdout.write("\n")
+```
+
+- [ ] **Step 2: End-to-end smoke (tiny, live)**
+
+Run:
+```bash
+cd skills/gitlink-spark/scripts && PYTHONUTF8=1 PYTHONIOENCODING=utf-8 python spark.py --field "graph neural network" --max-papers 3 --gap-type theory > _smoke.json 2>&1
+PYTHONUTF8=1 python -c "
+import json
+d=json.load(open('_smoke.json',encoding='utf-8'))
+print('field:', d['field'])
+print('papers:', len(d['papers']), '| gitlink_repos:', len(d['gitlink_repos']))
+print('github_counts keys:', len(d['github_counts']))
+g=d['github_counts']
+for k,v in list(g.items())[:1]: print('  sample gh:', k[:30], '-> total', v.get('total_count'))
+"
+rm -f _smoke.json
+```
+Expected: papers=3, gitlink_repos≥1, github_counts has 3 entries with real total_count ints. Takes ~25s (3 GitHub calls × 7s throttle). If GitHub 403 rate-limit, set `GITHUB_TOKEN` env or wait 60s.
+
+- [ ] **Step 3: Re-run unit tests (regression)**
+
+Run: `cd skills/gitlink-spark/scripts && python test_spark.py`
+Expected: `ALL TESTS PASSED`
+
+- [ ] **Step 4: Commit**
+
+```bash
+git add skills/gitlink-spark/scripts/spark.py
+git commit -m "feat(spark): fetch_github_count + main() 编排 → JSON（端到端 smoke 通过）"
+```
+
+---
+
+## Task 8: REFERENCE.md — gap taxonomy detail + GitHub tiers + LLM prompt + data sources
+
+**Files:**
+- Create: `skills/gitlink-spark/REFERENCE.md`
+
+- [ ] **Step 1: Create REFERENCE.md**
+
+Content (exact):
+
+```markdown
+# gitlink-spark 参考文档
+
+> SKILL.md 的深度参考：缺口分类细则、GitHub 阈值、LLM prompt 模板、数据源实测、诚实边界。
+
+## 一、缺口分类法细则
+
+### 类型 A 有理论无实现
+- 输入：arXiv 论文方法 M（title + method_keywords）
+- GitLink 侧：`search +repos -k <M>` 命中数（0 或极少，如 ≤2）
+- GitHub 侧：`total_count`（按 §二阈值分级）
+- 判定为"缺口"条件：GitLink ≤2 **且** GitHub < 50（全球稀缺或新兴）
+
+### 类型 B 有需求无解答
+- 输入：领域仓库的 open issue（`issue +list`，排除关闭）
+- LLM 筛"研究性痛点"：含性能/可扩展性/新场景/新数据集，排除安装报错/使用咨询
+- 判定：GitLink 无现成实现解此痛点 **且** GitHub 无成熟开源方案
+
+## 二、GitHub 全球对照阈值
+
+| total_count | 分级 | 报告 |
+|---|---|---|
+| < 10 | 全球稀缺 | 高价值缺口 |
+| 10–50 | 新兴 | 中等缺口 |
+| ≥ 50 | 已成熟 | **不报为空白**，列入"已诚实排除" |
+
+spark.py 缓存 GitHub 结果（按 query key），避免重复调用。
+
+## 三、LLM 缺口匹配 prompt 模板
+
+```
+你是科研机会发现助手。下面是 spark.py 抓取的真实数据（JSON）。
+请跨"arXiv 论文 × GitLink 仓库/issues × GitHub 全球计数"找出语义缺口，输出机会报告。
+
+规则：
+1. 只输出可溯源到下列数据的缺口；每张缺口卡带"实证三件套"。
+2. 类型A（理论无实现）：论文 M 的 GitLink 命中≤2 且 GitHub total_count<50 才报；
+   GitHub ≥50 的论文列入"已诚实排除"，不报为空白。
+3. 类型B（需求无解答）：只挑研究性痛点 issue，排除使用/安装类。
+4. 每张卡给一句"机会建议"（主观），但证据必须客观可查。
+5. 宁可少报，不误报。
+
+数据：
+{spark.py 的 JSON}
+```
+
+## 四、数据源实测结论（2026-07-01）
+
+| 源 | 状态 | 备注 |
+|---|---|---|
+| arXiv API | ✅ 必须 HTTPS | HTTP 被沙箱阻断返回 0 字节 |
+| gitlink-cli search +repos | ✅ | 用 identifier/关键词 |
+| gitlink-cli issue +list | ✅ | 逐仓库，绕开 search+issues |
+| gitlink-cli search +issues | ❌ 返回 HTML | 不可用，勿用 |
+| GitHub Search API | ✅ | 未认证 10/min；GITHUB_TOKEN 提额 |
+| OpenAlex | ⚠ 间歇 503 | best-effort 富集，降级跳过 |
+
+## 五、诚实边界
+
+1. **GitLink 覆盖薄**：缺口卡明确标 "GitLink 0 / GitHub N"；GitHub ≥50 不报为空白。
+2. LLM 缺口必须可溯源实证三件套，否则丢弃。
+3. arXiv 仅覆盖 CS/物理等，报告标注学科范围。
+4. GitHub 未认证 10/min：spark.py sleep 7s + 缓存；建议 demo 设 GITHUB_TOKEN。
+5. "机会建议"为主观启发，标注"需研究者自行判断"。
+```
+
+- [ ] **Step 2: Verify**
+
+Run: `grep -c -e "类型 A 有理论无实现" -e "GitHub ≥50 不报为空白" -e "search +issues.*HTML" -e "LLM 缺口匹配 prompt" skills/gitlink-spark/REFERENCE.md`
+Expected: `4`
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-spark/REFERENCE.md
+git commit -m "feat(spark): REFERENCE（缺口分类法+GitHub阈值+LLM prompt+数据源实测）"
+```
+
+---
+
+## Task 9: Live GNN run — produce 2-3 gap cards + ≥1 kickoff
+
+**Files:**
+- (no committed code; produces real data feeding Task 10's examples doc)
+
+- [ ] **Step 1: Run spark.py on GNN (full)**
+
+```bash
+cd skills/gitlink-spark/scripts
+PYTHONUTF8=1 PYTHONIOENCODING=utf-8 python spark.py --field "graph neural network" --max-papers 8 --gap-type both > _gnn.json 2>&1
+PYTHONUTF8=1 python -c "
+import json
+d=json.load(open('_gnn.json',encoding='utf-8'))
+print('papers:',len(d['papers']),'| repos:',len(d['gitlink_repos']),'| issues:',len(d['gitlink_issues']))
+# 列出 GitHub 稀缺(<10) 的论文 = 理论缺口候选
+for p in d['papers']:
+    q=p['title'][:60]; gh=d['github_counts'].get(q,{})
+    tc=gh.get('total_count')
+    if tc is not None and tc < 10:
+        print('  THEORY gap cand:', p['arxiv_id'], '|', p['title'][:45], '| GitHub', tc)
+"
+```
+Expected: papers=8, repos≥1, and ≥1 THEORY gap candidate (GitHub <10). Note the candidate arxiv_ids + GitHub counts. Keep `_gnn.json` for the examples doc.
+
+- [ ] **Step 2: LLM-match gap cards from _gnn.json**
+
+Following REFERENCE §三 prompt, produce the opportunity report from `_gnn.json`:
+- ≥1 "理论无实现" gap card (from Step 1 candidates) — with evidence triple
+- ≥1 "需求无解答" gap card (from gitlink_issues, if any research-pain issue; if issues empty/none research-y, note honestly and lean on theory gaps + lower the demand bar OR widen field keyword)
+- ≥1 "已诚实排除" entry (a paper with GitHub ≥50)
+Save the report text (will go into Task 10 examples). If no demand-side issue exists, be honest: report 2 theory gaps + 1 排除, note demand-side sparse for GNN on GitLink.
+
+- [ ] **Step 3: Kickoff (fork + issue) on ONE theory gap**
+
+Pick the best theory gap (GitHub <10). Fork its GitHub top repo's nearest GitLink equivalent OR the GitHub top repo isn't forkable via gitlink-cli (cross-platform) — instead: if a GitLink baseline exists, `gitlink-cli repo +fork` it; else create a todo issue on an existing GitLink GNN repo (e.g. GraphGallery) describing the reproduction plan with the paper's pseudocode.
+
+```bash
+# 若有 GitLink 基准仓库，fork 它；否则在 leejt/GraphGallery 开个复现 todo issue
+gitlink-cli issue +create --owner leejt --repo GraphGallery \
+  --title "Reproduction todo: <paper title> (gitlink-spark 机会)" \
+  --body "<论文 arxiv 链接 + Algorithm 伪代码摘要 + 机会报告卡>"
+```
+Capture the issue URL/number. If write fails (no permission on leejt/GraphGallery), fall back: create the issue on your own fork (fork first) OR output the todo text for manual creation. Note the actual outcome.
+
+- [ ] **Step 4: No commit (data-gathering)** — proceed to Task 10.
+
+---
+
+## Task 10: examples/spark-图神经网络.md — real walkthrough
+
+**Files:**
+- Create: `skills/gitlink-spark/examples/spark-图神经网络.md`
+
+- [ ] **Step 1: Write the real walkthrough using Task 9 outputs**
+
+Content skeleton (fill with REAL data from `_gnn.json` + Task 9 report/kickoff — no placeholders):
+
+```markdown
+# 示例：gitlink-spark GNN 缺口挖掘（真实数据）
+
+> 基于 `spark.py --field "graph neural network" --max-papers 8` 于 2026-07-XX 实跑。
+
+## 数据采集（真实）
+- arXiv 论文：N 篇（近 90 天）
+- GitLink 仓库：M 个（含 <列举>）
+- GitLink open issues：K 条
+- GitHub 全球对照：8 个查询
+
+## 机会报告（真实全文）
+<贴 Task 9 §2 的报告全文，含 ≥1 理论缺口 + ≥1 排除，每张带实证三件套>
+
+## 起跑（真实）
+<fork/issue 链接或降级说明>
+
+## 关键结论
+- 三源融合真实可跑
+- GitHub 阈值生效（≥50 不报为空白）
+- 每条缺口可溯源到 spark.py JSON
+```
+
+- [ ] **Step 2: Verify no placeholders**
+
+Run: `grep -E "TBD|TODO|<论文|<列举|<贴 Task" skills/gitlink-spark/examples/spark-图神经网络.md`
+Expected: no matches. If any, fill from Task 9 outputs.
+
+- [ ] **Step 3: Clean temp + commit**
+
+```bash
+rm -f skills/gitlink-spark/scripts/_gnn.json
+git add skills/gitlink-spark/examples/spark-图神经网络.md
+git commit -m "feat(spark): examples GNN 真实走查（2-3缺口卡+起跑）"
+```
+
+---
+
+## Task 11: README + workflow link + register + acceptance
+
+**Files:**
+- Modify: `skills/README.md`
+- Modify: `skills/gitlink-workflow/SKILL.md`
+
+- [ ] **Step 1: Register the skill**
+
+Run: `bash scripts/setup-skills.sh` then `ls ~/.claude/skills/ | grep gitlink-spark`
+Expected: `gitlink-spark` listed.
+
+- [ ] **Step 2: Add README row**
+
+In `skills/README.md` 智能 Skills table, after the gitlink-research-fair row add:
+
+```markdown
+| **gitlink-spark** | 文献-代码语义缺口挖掘机 | arXiv×GitLink×GitHub 三源挖"理论无实现/需求无解答"缺口，出机会报告，一键 fork+issue 起跑 |
+```
+
+Verify: `grep -c gitlink-spark skills/README.md` → `≥1`.
+
+- [ ] **Step 3: workflow cross-link**
+
+In `skills/gitlink-workflow/SKILL.md` 专项 Skill list, add:
+```markdown
+> - 科研机会发现（缺口挖掘） → [`../gitlink-spark/SKILL.md`](../gitlink-spark/SKILL.md)
+```
+
+- [ ] **Step 4: Acceptance (spec §15)**
+
+Verify:
+- [ ] 四件套齐全（SKILL.md + spark.py + REFERENCE.md + examples/）
+- [ ] `python spark.py --field 图神经网络` 输出合法 JSON
+- [ ] GNN 跑出 ≥1 理论 + ≥1 demand（或诚实标注 demand 稀疏）缺口卡，三件套可查
+- [ ] GitHub 阈值生效（≥1 "已诚实排除"）
+- [ ] ≥1 理论缺口走完起跑
+- [ ] 每条缺口可溯源 spark.py JSON
+- [ ] REFERENCE 含分类法+数据源实测+诚实边界
+- [ ] README 登记
+
+- [ ] **Step 5: Commit + push (updates PR #5)**
+
+```bash
+git add skills/README.md skills/gitlink-workflow/SKILL.md
+git commit -m "docs(spark): README 登记 + workflow 链接 + 注册"
+git push myfork feat/gitlink-research-fair
+```

docs/superpowers/plans/2026-07-07-gitlink-research-fair-v2.md

@@ -0,0 +1,759 @@
+# gitlink-research-fair v2 Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Upgrade `gitlink-research-fair` from "5-axis FAIR card" to **科研软件 X 光 (Research Software X-Ray)** — a runnable `fair.py` extracts a real research profile (paper/datasets/repro/citation), LLM renders a varied per-repo report (verdict line + real Mermaid knowledge graph + repo-specific findings) with optional prescription.
+
+**Architecture:** `scripts/fair.py` = stdlib-only deterministic extractor (README regex + file scan → research-profile JSON). `SKILL.md` orchestrates the LLM: 4-dimension verdicts (each citing fair.py evidence) → real knowledge graph → hybrid report → optional fork+PR prescription. Pure extractors are unit-tested; Feature_Critic + Edge live runs are integration validation.
+
+**Tech Stack:** Python 3.11 stdlib only (argparse/json/subprocess/re). gitlink-cli (file +get/+list, repo +info, commit +list). Mermaid for the graph.
+
+**Source of truth:** `docs/superpowers/specs/2026-07-07-gitlink-research-fair-v2-design.md` (read it first).
+
+---
+
+## Environment & Gotchas (engineer must know)
+
+- **Branch:** `feat/gitlink-research-fair` (v2 升级与 v1/spark 同 PR #5). Commit only your own files; leave pre-existing `D README_TASKB.md` / `D gitlink-cli.exe` / `?? dist/` / `?? _edge_prescription/` / `?? report-cards/` untouched.
+- **Encoding:** python touching gitlink-cli output MUST run `PYTHONUTF8=1 PYTHONIOENCODING=utf-8`. Never inline Chinese in `python -c` — write a `.py` file.
+- **`file +list` returns `data` as a stringified JSON** → `json.loads` it (known structure, see Task 3).
+- **`file +get` content lives at `data.entries.content`** (plain text, not base64).
+- **`--repo` uses identifier** (ASCII slug, e.g. `Feature_Critic`), not Chinese display name.
+- **stdlib only** — no pip. Tests run via `python test_fair.py` (assert-based, no pytest).
+- **v1 to remove:** the old `examples/research-fair-workflow.md` (songhui18 v1 report card) is replaced by v2 examples (Task 8 `git rm`s it).
+
+---
+
+## File Structure
+
+| File | Responsibility |
+|------|----------------|
+| `skills/gitlink-research-fair/scripts/fair.py` | Stdlib extractor: README regex + file scan → research-profile JSON. Pure extractors + gitlink-cli fetchers + main() |
+| `skills/gitlink-research-fair/scripts/test_fair.py` | Assert-based unit tests for pure extractors (extract_paper, extract_datasets, assess_repro, assess_citation, extract_methods_frameworks) |
+| `skills/gitlink-research-fair/SKILL.md` | REWRITE: X-ray pipeline, 4-dim verdict table, hybrid report template, real KG section, prescription guardrails, degradation table |
+| `skills/gitlink-research-fair/REFERENCE.md` | REWRITE: 4-dim verdict rules, real KG schema, extraction rules, data-source findings, FAIR4RS anchor |
+| `skills/gitlink-research-fair/examples/feature-critic-xray.md` | Real Feature_Critic X-ray (live-demo script) |
+| `skills/gitlink-research-fair/examples/edge-xray.md` | Real Edge X-ray (engine-class contrast: license-conflict finding) |
+| `skills/README.md` + `skills/gitlink-workflow/SKILL.md` | Update fair description → "科研软件 X 光" |
+
+---
+
+## Task 1: fair.py — extract_paper + extract_datasets + tests (TDD)
+
+**Files:**
+- Create: `skills/gitlink-research-fair/scripts/fair.py`
+- Create: `skills/gitlink-research-fair/scripts/test_fair.py`
+
+- [ ] **Step 1: Write the failing test**
+
+`test_fair.py` (exact):
+
+```python
+# -*- coding: utf-8 -*-
+"""Assert-based unit tests for fair.py pure extractors. Run: python test_fair.py"""
+import sys, os
+sys.path.insert(0, os.path.dirname(__file__))
+from fair import extract_paper, extract_datasets, assess_repro, assess_citation, extract_methods_frameworks
+
+SAMPLE_README = """# Feature_Critic
+Demo code for 'Feature-Critic Networks for Heterogeneous Domain Generalisation'.
+This paper is located at https://arxiv.org/abs/1901.11448 and will appear in ICML 2019.
+Evaluated on PACS and Visual Decathlon.
+
+@inproceedings{li2019feature,
+  title={Feature-Critic Networks},
+  booktitle={ICML}}
+"""
+
+SAMPLE_FILES = ["README.md", "main_Feature_Critic.py", "main_baseline.py", "model_PACS.py",
+                "alexnet.py", "resnet.py", "vggnet.py", "data_gen_PACS.py", "get_model_dataset.sh", "utils.py"]
+
+def test_extract_paper():
+    p = extract_paper(SAMPLE_README)
+    assert p["arxiv_id"] == "1901.11448", p["arxiv_id"]
+    assert p["arxiv_url"] == "https://arxiv.org/abs/1901.11448"
+    assert p["venue"] == "ICML"
+    assert p["in_readme"] is True
+    print("test_extract_paper OK")
+
+def test_extract_paper_none():
+    p = extract_paper("# Hello\nA normal project with no paper.")
+    assert p["in_readme"] in (False, True)  # title-only may set in_readme; arxiv must be None
+    assert p["arxiv_id"] is None
+    print("test_extract_paper_none OK")
+
+def test_extract_datasets():
+    ds = extract_datasets(SAMPLE_README, SAMPLE_FILES)
+    names = [d["name"] for d in ds]
+    assert "PACS" in names and "Visual Decathlon" in names
+    pacs = [d for d in ds if d["name"] == "PACS"][0]
+    assert pacs["download_script"] and "data_gen_PACS.py" in pacs["download_script"]
+    print("test_extract_datasets OK")
+
+if __name__ == "__main__":
+    test_extract_paper(); test_extract_paper_none(); test_extract_datasets()
+    print("PART 1 OK (run all after Task 2)")
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run: `cd skills/gitlink-research-fair/scripts && python test_fair.py`
+Expected: FAIL with `ModuleNotFoundError: No module named 'fair'`
+
+- [ ] **Step 3: Write fair.py with extract_paper + extract_datasets (+ stubs for Task 2 functions so import works)**
+
+`fair.py` (exact):
+
+```python
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""gitlink-research-fair v2: research software X-ray. Extract a research profile from a GitLink repo. Stdlib only."""
+import argparse, json, os, sys, subprocess, re
+
+_ARXIV_PATS = [
+    r'https?://arxiv\.org/(?:abs|pdf)/(\d{4}\.\d{4,5})',
+    r'arXiv:(\d{4}\.\d{4,5})',
+    r'\b(\d{4}\.\d{4,5})\b',
+]
+_DOI_PAT = r'10\.\d{4,9}/\S+'
+_VENUES = ["ICML", "NeurIPS", "NIPS", "ICLR", "CVPR", "ICCV", "ECCV", "ACL", "EMNLP",
+           "NAACL", "KDD", "WWW", "AAAI", "IJCAI", "SIGGRAPH", "Nature", "Science"]
+_KNOWN_DATASETS = ["Visual Decathlon", "PACS", "ImageNet", "CIFAR-10", "CIFAR-100", "CIFAR",
+                   "Cora", "Citeseer", "Pubmed", "MNIST", "COCO", "QM9", "ZINC", "OGB", "ogbn",
+                   "Wikipedia", "PPI", "Reddit", "Amazon", "Yelp", "MUTAG"]
+
+def extract_paper(readme):
+    """Extract paper provenance (arxiv/doi/title/venue) from README text."""
+    if not readme:
+        return {"in_readme": False, "arxiv_id": None, "arxiv_url": None, "doi": None,
+                "title": None, "authors": [], "venue": None}
+    arxiv_id = arxiv_url = None
+    for pat in _ARXIV_PATS:
+        m = re.search(pat, readme)
+        if m:
+            arxiv_id = m.group(1); arxiv_url = f"https://arxiv.org/abs/{m.group(1)}"; break
+    doi = None
+    m = re.search(_DOI_PAT, readme)
+    if m:
+        doi = m.group(0).rstrip(").,;]")
+    venue = None
+    for v in _VENUES:
+        if re.search(rf"\b{re.escape(v)}\b", readme):
+            venue = v; break
+    title = None
+    m2 = (re.search(r"[Cc]ode (?:for|of)\s+'([^']+)'", readme)
+          or re.search(r'[Cc]ode (?:for|of)\s+"([^"]+)"', readme)
+          or re.search(r"^\s*#\s+(.+)$", readme, re.M))
+    if m2:
+        title = m2.group(1).strip()
+    return {"in_readme": bool(arxiv_id or doi or title), "arxiv_id": arxiv_id,
+            "arxiv_url": arxiv_url, "doi": doi, "title": title, "authors": [], "venue": venue}
+
+def extract_datasets(readme, files):
+    """Identify referenced datasets (known-name match + data scripts)."""
+    text = readme or ""
+    found = []
+    for ds in _KNOWN_DATASETS:
+        if re.search(rf"\b{re.escape(ds)}\b", text, re.I):
+            found.append(ds)
+    scripts = [f for f in files if any(k in (f or "").lower()
+                                        for k in ["data_gen", "get_data", "download", "prepare_data", "data_load"])]
+    return [{"name": ds, "evidence": "mentioned in README",
+             "download_script": scripts[:2] or None, "license": None} for ds in found]
+
+def assess_repro(files, readme):  # implemented in Task 2
+    return {"deps_files": [], "deps_pinned": False, "entry_points": [], "expected_results": False, "env_spec": False}
+
+def assess_citation(files, readme):  # implemented in Task 2
+    return {"cff": False, "codemeta": False, "zenodo": False, "readme_bibtex": None}
+
+def extract_methods_frameworks(files, readme):  # implemented in Task 2
+    return {"methods": [], "frameworks": []}
+
+def main():  # implemented in Task 3
+    pass
+
+if __name__ == "__main__":
+    main()
+```
+
+- [ ] **Step 4: Run test to verify Task-1 tests pass**
+
+Run: `cd skills/gitlink-research-fair/scripts && python test_fair.py`
+Expected: `test_extract_paper OK` / `test_extract_paper_none OK` / `test_extract_datasets OK` / `PART 1 OK`
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add skills/gitlink-research-fair/scripts/fair.py skills/gitlink-research-fair/scripts/test_fair.py
+git commit -m "feat(fair-v2): fair.py extract_paper + extract_datasets + test（TDD）"
+```
+
+---
+
+## Task 2: fair.py — assess_repro + assess_citation + extract_methods_frameworks (real impl + tests)
+
+**Files:**
+- Modify: `skills/gitlink-research-fair/scripts/fair.py` (replace the 3 stubs)
+- Modify: `skills/gitlink-research-fair/scripts/test_fair.py` (add tests)
+
+- [ ] **Step 1: Add tests for the 3 functions**
+
+Append to `test_fair.py` (before the `if __name__` block):
+
+```python
+def test_assess_repro():
+    files = ["README.md", "main_Feature_Critic.py", "requirements.txt", "model_PACS.py"]
+    r = assess_repro(files, SAMPLE_README)
+    assert "requirements.txt" in r["deps_files"]
+    assert r["deps_pinned"] is True
+    assert "main_Feature_Critic.py" in r["entry_points"]
+    assert r["expected_results"] is False  # SAMPLE_README has no accuracy/results table
+    print("test_assess_repro OK")
+
+def test_assess_citation():
+    c = assess_citation(SAMPLE_FILES, SAMPLE_README)
+    assert c["cff"] is False and c["codemeta"] is False
+    assert c["readme_bibtex"] and "@inproceedings" in c["readme_bibtex"]
+    print("test_assess_citation OK")
+
+def test_extract_methods_frameworks():
+    mf = extract_methods_frameworks(SAMPLE_FILES, SAMPLE_README)
+    assert "domain generalisation" in mf["methods"], mf["methods"]  # SAMPLE_README 提到 Domain Generalisation
+    assert isinstance(mf["frameworks"], list)
+    print("test_extract_methods_frameworks OK")
+```
+
+And replace the `if __name__ == "__main__":` block with:
+
+```python
+if __name__ == "__main__":
+    test_extract_paper(); test_extract_paper_none(); test_extract_datasets()
+    test_assess_repro(); test_assess_citation(); test_extract_methods_frameworks()
+    print("ALL TESTS PASSED")
+```
+
+- [ ] **Step 2: Run test to verify the 3 new ones fail**
+
+Run: `cd skills/gitlink-research-fair/scripts && python test_fair.py`
+Expected: `test_assess_repro OK` may print then FAIL on assert (stubs return empty deps_files → `"requirements.txt" in []` is False → assert fails). Confirms stubs need replacing.
+
+- [ ] **Step 3: Replace the 3 stubs with real implementations**
+
+Replace `def assess_repro(...) ...` / `def assess_citation(...) ...` / `def extract_methods_frameworks(...) ...` with:
+
+```python
+def assess_repro(files, readme):
+    """Static reproducibility readiness: deps + entry + env + expected results."""
+    name_set = {(f or "") for f in files}
+    deps_candidates = ["requirements.txt", "environment.yml", "go.mod", "package.json",
+                       "Dockerfile", "setup.py", "pyproject.toml"]
+    deps_files = [f for f in deps_candidates if f in name_set]
+    entry_points = sorted([f for f in name_set if re.match(r"(main|train|run|demo)_?\w*\.py$", f, re.I)])
+    env_spec = any(f in ("Dockerfile", "environment.yml") for f in deps_files)
+    expected = bool(re.search(r"(accuracy|f1\b|bleu|rouge|results?\s*(table|in section)|table\s*\d)",
+                              readme or "", re.I))
+    return {"deps_files": deps_files, "deps_pinned": bool(deps_files),
+            "entry_points": entry_points[:5], "expected_results": expected, "env_spec": env_spec}
+
+def assess_citation(files, readme):
+    """Citation readiness: CITATION.cff / codemeta / zenodo + README bibtex."""
+    name_set = {(f or "") for f in files}
+    m = re.search(r"@(inproceedings|article|misc|book)\{[^}]+\}", readme or "", re.S | re.I)
+    return {"cff": "CITATION.cff" in name_set,
+            "codemeta": "codemeta.json" in name_set,
+            "zenodo": ".zenodo.json" in name_set,
+            "readme_bibtex": (m.group(0)[:200] if m else None)}
+
+def extract_methods_frameworks(files, readme):
+    """Infer methods + frameworks from filenames + README."""
+    text = " ".join(files) + " " + (readme or "")
+    frameworks = []
+    if re.search(r"\b(torch|pytorch|nn\.module)\b", text, re.I): frameworks.append("PyTorch")
+    if re.search(r"\b(tensorflow|tf\.|keras)\b", text, re.I): frameworks.append("TensorFlow")
+    if re.search(r"\b(jax|flax|haiku)\b", text, re.I): frameworks.append("JAX")
+    if re.search(r"\b(sklearn|scikit-learn)\b", text, re.I): frameworks.append("scikit-learn")
+    methods = []
+    for kw in ["attention", "transformer", "contrastive", "meta-learning", "federated",
+               "graph", "convolution", "resnet", "gan", "diffusion", "reinforcement",
+               "domain generalisation", "domain generalization"]:
+        if re.search(rf"\b{kw}", text, re.I):
+            methods.append(kw)
+    return {"methods": methods[:6],
+            "frameworks": frameworks or ["unknown (infer from filenames; verify imports)"]}
+```
+
+- [ ] **Step 4: Run all tests — expect ALL PASS**
+
+Run: `cd skills/gitlink-research-fair/scripts && python test_fair.py`
+Expected: `ALL TESTS PASSED` (6 tests)
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add skills/gitlink-research-fair/scripts/fair.py skills/gitlink-research-fair/scripts/test_fair.py
+git commit -m "feat(fair-v2): assess_repro + assess_citation + extract_methods_frameworks（6 测试全过）"
+```
+
+---
+
+## Task 3: fair.py — gitlink-cli fetchers + main + end-to-end smoke
+
+**Files:**
+- Modify: `skills/gitlink-research-fair/scripts/fair.py` (add network layer + replace main stub)
+
+- [ ] **Step 1: Add gitlink-cli fetchers + implement main()**
+
+Insert before `def main():` and replace the `def main(): pass` stub:
+
+```python
+def _gitlink(*args):
+    """Run gitlink-cli with json output; return parsed dict (UTF-8 safe)."""
+    r = subprocess.run(["gitlink-cli"] + list(args) + ["--format", "json"],
+                       capture_output=True, text=True, encoding="utf-8", errors="replace", timeout=60)
+    raw = r.stdout
+    i = raw.find("{")
+    return json.loads(raw[i:]) if i >= 0 else {}
+
+def fetch_readme(owner, repo):
+    d = _gitlink("file", "+get", "--owner", owner, "--repo", repo, "--path", "README.md")
+    ent = (d.get("data", {}) or {}).get("entries", {}) or {}
+    return ent.get("content", "") if isinstance(ent, dict) else ""
+
+def fetch_file_list(owner, repo):
+    d = _gitlink("file", "+list", "--owner", owner, "--repo", repo)
+    fd = d.get("data", "[]")
+    if isinstance(fd, str):
+        fd = json.loads(fd)
+    return [f.get("name") for f in fd if isinstance(f, dict)] if isinstance(fd, list) else []
+
+def fetch_repo_meta(owner, repo):
+    info = _gitlink("repo", "+info", "--owner", owner, "--repo", repo)
+    comm = _gitlink("commit", "+list", "--owner", owner, "--repo", repo, "--page", "1")
+    cd = comm.get("data", {})
+    cl = cd.get("commits") if isinstance(cd, dict) else None
+    head = (cl[0].get("sha") if cl and isinstance(cl, list) and cl else None)
+    d = info.get("data", {}) or {}
+    return {"identifier": d.get("identifier"), "license_id": d.get("license_id"),
+            "has_dataset": d.get("has_dataset"), "head_sha": head}
+
+def main():
+    ap = argparse.ArgumentParser(description="gitlink-research-fair v2: research software X-ray")
+    ap.add_argument("--owner", required=True)
+    ap.add_argument("--repo", required=True)
+    args = ap.parse_args()
+
+    readme = fetch_readme(args.owner, args.repo)
+    files = fetch_file_list(args.owner, args.repo)
+    meta = fetch_repo_meta(args.owner, args.repo)
+    mf = extract_methods_frameworks(files, readme)
+    profile = {
+        "repo": f"{args.owner}/{args.repo}",
+        "head_sha": meta.get("head_sha"),
+        "paper": extract_paper(readme),
+        "datasets": extract_datasets(readme, files),
+        "repro": assess_repro(files, readme),
+        "citation": assess_citation(files, readme),
+        "methods": mf["methods"],
+        "frameworks": mf["frameworks"],
+        "license": {"file": any("LICENSE" in (f or "") for f in files),
+                    "license_id": meta.get("license_id")},
+        "files_count": len(files),
+    }
+    json.dump(profile, sys.stdout, ensure_ascii=False, indent=2)
+    sys.stdout.write("\n")
+```
+
+- [ ] **Step 2: End-to-end smoke (live, Feature_Critic)**
+
+Run:
+```bash
+cd skills/gitlink-research-fair/scripts
+PYTHONUTF8=1 PYTHONIOENCODING=utf-8 python fair.py --owner liyiying10 --repo Feature_Critic > _xray.json 2>&1
+PYTHONUTF8=1 python -c "
+import json
+d=json.load(open('_xray.json',encoding='utf-8'))
+print('repo:', d['repo'], '| head:', (d.get('head_sha') or '')[:7])
+print('paper arxiv:', d['paper']['arxiv_id'], '| venue:', d['paper']['venue'], '| in_readme:', d['paper']['in_readme'])
+print('datasets:', [x['name'] for x in d['datasets']])
+print('repro deps:', d['repro']['deps_files'], '| entry:', d['repro']['entry_points'], '| env:', d['repro']['env_spec'])
+print('citation cff:', d['citation']['cff'], '| bibtex?', bool(d['citation']['readme_bibtex']))
+print('files:', d['files_count'])
+"
+rm -f _xray.json
+```
+Expected: arxiv `1901.11448` + venue `ICML` + datasets incl `PACS`/`Visual Decathlon` + entry `main_Feature_Critic.py` + cff False + bibtex True + files 15. If arxiv None, inspect README (the repo may have changed; adapt regex). If `file +list` parse fails, confirm data is stringified-JSON and json.loads handles it.
+
+- [ ] **Step 3: Re-run unit tests (regression)**
+
+Run: `cd skills/gitlink-research-fair/scripts && python test_fair.py`
+Expected: `ALL TESTS PASSED`
+
+- [ ] **Step 4: Commit**
+
+```bash
+git add skills/gitlink-research-fair/scripts/fair.py
+git commit -m "feat(fair-v2): gitlink-cli fetchers + main() 编排 → 科研画像 JSON（Feature_Critic smoke 通过）"
+```
+
+---
+
+## Task 4: Rewrite SKILL.md (X-ray pipeline + 4-dim + hybrid report + real KG + prescription + degradation)
+
+**Files:**
+- Modify (overwrite): `skills/gitlink-research-fair/SKILL.md`
+
+- [ ] **Step 1: Overwrite SKILL.md with the v2 content**
+
+New `SKILL.md` (exact — this replaces the v1 5-axis content entirely):
+
+```markdown
+---
+name: gitlink-research-fair
+version: 2.0.0
+description: "科研软件 X 光：用 fair.py 真抽取 GitLink 科研仓库的论文/数据/复现/引用画像，LLM 四维裁决，输出含真科研图谱与特有关键发现的洞察报告，可选处方 PR。当用户需要深挖科研仓库的科研产物、评估可复现/可引用性时触发。"
+metadata:
+  requires:
+    bins: ["gitlink-cli"]
+  cliHelp: "python skills/gitlink-research-fair/scripts/fair.py --help"
+---
+
+# gitlink-research-fair v2（科研软件 X 光）
+
+**CRITICAL — 开始前必须先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)，其中包含认证、权限处理和 API 注意事项。**
+**CRITICAL — 裁决由 LLM 做出，但每条必须引用 `fair.py` 抽到的实证（arxiv id / 文件名 / deps 状态）；无实证的判断丢弃。**
+**CRITICAL — 处方（开 PR）默认预览确认；绝不自动 merge、绝不 force-push、绝不碰原仓库。**
+**CRITICAL — GitLink 操作只能用 `gitlink-cli`。禁止用 `gh`。**
+
+> **前置条件：** 先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)；四维裁决细则、真 KG schema、抽取规则见 [`REFERENCE.md`](REFERENCE.md)。
+
+## 概述
+
+**科研软件 X 光**：`scripts/fair.py` 真抽取仓库内容（README/文件/依赖）→ 科研画像 JSON；LLM 对 4 个科研专属维度裁决（论文溯源/数据链/复现就绪/引用就绪），渲染**每 repo 特有的洞察报告**（裁决总览 + 真科研图谱 + 关键发现），可选处方 PR。与 `gitlink-health`（项目过程健康）正交，与 `gitlink-spark`（跨仓挖缺口）互补——本 skill **单仓深挖科研产物**。
+
+## 命令接口
+
+```bash
+python skills/gitlink-research-fair/scripts/fair.py --owner <owner> --repo <identifier>
+# → stdout: 科研画像 JSON {paper, datasets, repro, citation, methods, frameworks, license, head_sha, files_count}
+```
+
+skill 约定参数（非 CLI flag）：
+
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--owner/--repo` | 自动从 cwd 解析 | 目标科研仓库（用 identifier） |
+| `--auto` | 关 | 跳过预览直接开处方 PR（仍受护栏） |
+| `--no-fork` | 关 | 只出 X 光报告，不开 PR |
+| `--refresh` | 关 | 即使有旧哨兵也重评 |
+
+## 管道
+
+### ① 抽取（fair.py，确定性）
+`fetch_readme`（file +get）+ `fetch_file_list`（file +list）+ `fetch_repo_meta`（repo +info + commit +list 取 HEAD sha）→ `extract_paper` / `extract_datasets` / `assess_repro` / `assess_citation` / `extract_methods_frameworks` → 科研画像 JSON
+
+### ② 四维裁决（LLM，读 JSON）
+论文溯源 / 数据链 / 复现就绪 / 引用就绪。每维 `✅/⚠️/❌` + **引用画像字段的具体证据**。规则见 [`REFERENCE.md`](REFERENCE.md)。
+
+### ③ 真科研图谱（从画像生成 Mermaid）
+节点 `Paper↔Method↔Code↔Dataset↔Framework↔Citation`，带状态色（✅绿/⚠️黄/❌红）。schema 见 REFERENCE。
+
+### ④ 渲染 X 光报告（hero）并落盘
+裁决一行 + 真图谱 + **本 repo 特有关键发现** + 处方摘要 + 双裁决证书。**始终保存** `report-cards/<owner>-<repo>-xray.md`（绝不只在终端）。
+
+### ⑤ 处方（可选）
+对 ❌/⚠️ 项生成 CITATION.cff（从 README 抽的引用）/ requirements.txt（从 import 扫）/ Dockerfile → fork → PR（默认预览）。
+
+## 四维裁决（速览，细则见 REFERENCE）
+
+| 维度 | ✅ | ⚠️ | ❌ |
+|------|---|----|----|
+| 论文溯源 | arxiv/DOI + 元数据全 | 仅 README 文字，无稳定链接 | 无论文线索 |
+| 数据链 | 命名 + 下载脚本 + license | 命名但无脚本/无 license | 未提及数据集 |
+| 复现就绪 | 依赖锁+入口+环境+期望结果齐全 | 有入口但缺依赖锁/环境/期望结果 | 无入口/无依赖 |
+| 引用就绪 | CITATION.cff/codemeta + DOI + 版本 | 仅 README 引用文本 | 无引用信息 |
+
+## 报告格式（混合主视觉，hero）
+
+````markdown
+🔬 **科研软件 X 光 — <owner>/<repo>**
+
+═══════════════════════════════════════
+论文溯源 <V> ｜ 数据链 <V> ｜ 复现就绪 <V> ｜ 引用就绪 <V>
+═══════════════════════════════════════
+
+### 🧬 真科研图谱
+```mermaid
+graph LR
+  P[<paper venue+arxiv>]:::ok -->|proposes| M[<method>]
+  M -->|implements| C[<entry file>]:::ok
+  C -->|uses| D[<datasets>]:::warn
+  C -->|depends| F[<framework>]:::warn
+  P -->|cited-via| Ci[<CITATION? or 无>]:::bad
+  classDef ok fill:#cfe,stroke:#3a3; classDef warn fill:#ffe,stroke:#cc3; classDef bad fill:#fee,stroke:#c33;
+```
+
+### 🔍 关键发现（本 repo 特有）
+- <LLM 从画像抽出的 ≥3 条具体发现，每条引用 fair.py 字段>
+
+### 🔧 处方（可选）
+- <对 ❌/⚠️ 项的修复建议>
+
+### 📜 双裁决证书
+复现就绪 <V> ｜ 引用就绪 <V> ｜ 锚定 commit `<sha>`
+
+---
+<!-- gitlink-research-fair v2 | repo:<owner>/<repo> | paper:<✅/⚠️/❌> | repro:<V> | cite:<V> | sha:<head> -->
+*由 gitlink-research-fair v2（科研软件 X 光）生成。*
+````
+
+## 处方闭环 + 护栏
+
+对 ❌/⚠️ 项生成修复：`CITATION.cff`（从 README 抽的引用文本构造）+ `requirements.txt`（从代码 import 扫）+ `Dockerfile`（模板）→ `repo +fork` → `pr +create`（PR body 带报告摘要）。**护栏**：默认预览；`--auto` 跳过但**永不 force-push、永不碰原仓库、永不自动 merge**；`--no-fork` 报告已落盘（④），不开 PR。
+
+## 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| README 读失败 | 降级用 file list + 元数据，标注"README 不可读，结论受限" |
+| `file +list` data 为字符串 | json.loads 解套 |
+| arxiv/DOI 抽不到 | 论文溯源判 ⚠️/❌，据实 |
+| 框架无法推断 | frameworks 标"未知"，不编造 |
+| `--repo` 用中文显示名 404 | 提示用 identifier |
+| fork/PR 失败 | 处方物料落本地，告知路径 |
+| 报告/PR 发布失败 | 报告已在 ④ 落盘，告知路径 |
+```
+
+- [ ] **Step 2: Verify the rewrite removed v1 content**
+
+Run: `grep -c -e "5 轴评分 Rubric" -e "Repro 可复现" -e "五轴雷达" skills/gitlink-research-fair/SKILL.md`
+Expected: `0` (v1 5-axis content gone). Then `grep -c -e "科研软件 X 光" -e "四维裁决" -e "真科研图谱" skills/gitlink-research-fair/SKILL.md` → expect `≥3`.
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/SKILL.md
+git commit -m "feat(fair-v2): 重写 SKILL.md（X光管道+四维裁决+混合报告+真图谱+处方），移除v1五轴"
+```
+
+---
+
+## Task 5: Rewrite REFERENCE.md (4-dim rules + KG schema + extraction rules + FAIR4RS anchor)
+
+**Files:**
+- Modify (overwrite): `skills/gitlink-research-fair/REFERENCE.md`
+
+- [ ] **Step 1: Overwrite REFERENCE.md with v2 content**
+
+New `REFERENCE.md` (exact — replaces v1):
+
+```markdown
+# gitlink-research-fair v2 参考文档（科研软件 X 光）
+
+> SKILL.md 的深度参考：四维裁决细则、真 KG schema、fair.py 抽取规则、数据源实测、学术锚。
+
+## 一、四维裁决细则
+
+每维 `✅/⚠️/❌`，**必须引用 fair.py 画像字段作证据**。
+
+### 论文溯源（F2/R1.2）
+- ✅ `paper.arxiv_id` 或 `paper.doi` 非空 + `paper.title`/`venue` 抽到
+- ⚠️ 仅 `paper.title` 抽到（"Code for..." 句式），无 arxiv/DOI
+- ❌ `paper.in_readme` = False
+
+### 数据链（FAIR 数据维度）
+- ✅ `datasets` 非空 + 至少一个有 `download_script` + 数据 license 可考
+- ⚠️ `datasets` 非空（命名）但无 download_script 或无 license
+- ❌ `datasets` 为空
+
+### 复现就绪（独立于 FAIR）
+- ✅ `repro.deps_pinned`=True + `entry_points` 非空 + `env_spec`=True + `expected_results`=True（四件齐全）
+- ⚠️ 有 `entry_points` 但缺依赖锁/环境/期望结果中任一
+- ❌ 无 `entry_points` 或无 `deps_files`
+
+### 引用就绪（R1.1/R2）
+- ✅ `citation.cff` 或 `citation.codemeta` 为 True + 有版本/DOI
+- ⚠️ 仅 `citation.readme_bibtex` 非空（README 有引用文本，无机器可读文件）
+- ❌ 三者皆 False
+
+## 二、真科研图谱 schema（Mermaid）
+
+节点：`Paper(venue+arxiv)` / `Method` / `Code(entry file)` / `Dataset` / `Framework` / `Citation`
+边：`Paper —proposes→ Method`、`Method —implements→ Code`、`Code —uses→ Dataset`、`Code —depends→ Framework`、`Paper —cited-via→ Citation`
+状态色（按对应维度裁决）：✅ `classDef ok fill:#cfe` / ⚠️ `classDef warn fill:#ffe` / ❌ `classDef bad fill:#fee`
+
+## 三、fair.py 抽取规则
+
+- **arxiv**：三路正则（arxiv URL / `arXiv:id` / 裸 `\d{4}.\d{4,5}`），取首个命中
+- **venue**：白名单（ICML/NeurIPS/ICLR/CVPR/ACL/...）正则
+- **datasets**：已知名白名单（PACS/Visual Decathlon/Cora/ImageNet/...）+ 数据脚本（data_gen/get_data/download）
+- **repro**：依赖文件名匹配（requirements/go.mod/environment.yml/Dockerfile/setup.py）+ 入口（main/train/run*.py）+ 期望结果（accuracy/f1/results table 正则）
+- **citation**：CITATION.cff/codemeta.json/.zenodo.json 文件存在 + README `@inproceedings/@article` bibtex
+- **frameworks**：torch/tensorflow/jax/sklearn 关键词（文件名+README），无则标"未知，verify imports"
+
+## 四、数据源实测（2026-07）
+
+| 源 | 状态 | 备注 |
+|---|---|---|
+| `gitlink-cli file +get` | ✅ | content 在 `data.entries.content`（纯文本） |
+| `gitlink-cli file +list` | ✅ | `data` 是字符串化 JSON，需 json.loads |
+| `gitlink-cli repo +info` | ✅ | license_id/identifier/has_dataset |
+| `gitlink-cli commit +list` | ✅ | HEAD sha |
+| OpenAlex | ❌ 已砍 | v1 弱环节（间歇 503），v2 不依赖 |
+
+## 五、学术锚（FAIR4RS）
+
+四维裁决对标 **FAIR4RS**（Barker et al. 2022, Nature Sci Data）：论文溯源→F2/R1.2、数据链→FAIR-Data、复现就绪→（独立轴，FAIR 必要非充分）、引用就绪→R1.1/R2。诚实声明：这是适配版评分（社区尚无认证级自动校验器），非官方认证。
+
+## 六、诚实边界
+
+1. fair.py 抽取覆盖度受 README 写法影响；非标准 README 可能漏（同时匹配多句式兜底）。
+2. 框架/方法为推断，标"推断"/"未知"，不肯定。
+3. 复现就绪是**静态**判断（依赖/入口/环境/期望结果四件套），不实际跑代码。
+4. v1 的 OpenAlex 溯源已砍（避免 503 弱环节）。
+```
+
+- [ ] **Step 2: Verify**
+
+Run: `grep -c -e "四维裁决细则" -e "真科研图谱 schema" -e "fair.py 抽取规则" -e "FAIR4RS" skills/gitlink-research-fair/REFERENCE.md`
+Expected: `4`. And `grep -c -e "5 轴" -e "OpenAlex 字段" skills/gitlink-research-fair/REFERENCE.md` → `0`.
+
+- [ ] **Step 3: Commit**
+
+```bash
+git add skills/gitlink-research-fair/REFERENCE.md
+git commit -m "feat(fair-v2): 重写 REFERENCE（四维裁决细则+真KG schema+抽取规则+FAIR4RS锚）"
+```
+
+---
+
+## Task 6: Live run — Feature_Critic X-ray (live-demo) + 落盘
+
+**Files:**
+- (produces `report-cards/liyiying10-Feature_Critic-xray.md`; data feeds Task 8 example)
+
+- [ ] **Step 1: Run fair.py on Feature_Critic**
+
+```bash
+cd skills/gitlink-research-fair/scripts
+PYTHONUTF8=1 PYTHONIOENCODING=utf-8 python fair.py --owner liyiying10 --repo Feature_Critic > _fc.json 2>&1
+PYTHONUTF8=1 python -c "
+import json
+d=json.load(open('_fc.json',encoding='utf-8'))
+assert d['paper']['arxiv_id']=='1901.11448', 'arxiv 抽取失败'
+assert d['paper']['venue']=='ICML'
+assert any(x['name']=='PACS' for x in d['datasets'])
+assert 'main_Feature_Critic.py' in d['repro']['entry_points']
+assert d['citation']['cff'] is False
+print('Feature_Critic 画像 OK | head:', (d.get('head_sha') or '')[:7])
+"
+```
+Expected: assertions pass. Keep `_fc.json` for the report.
+
+- [ ] **Step 2: LLM-render the X-ray report (controller)**
+
+Read `_fc.json`, apply REFERENCE §一 verdict rules, render the full hybrid report (verdict line + Mermaid graph + ≥3 repo-specific findings + prescription + dual-verdict certificate) per SKILL.md §报告格式. Expected verdicts: 论文溯源 ✅ / 数据链 ⚠️ / 复现就绪 ⚠️ / 引用就绪 ❌.
+
+- [ ] **Step 3: Save report to report-cards/**
+
+Save the rendered report to `report-cards/liyiying10-Feature_Critic-xray.md` (with sentinel). This is the live-demo artifact.
+
+- [ ] **Step 4: No commit (data-gathering)** — proceed to Task 8.
+
+---
+
+## Task 7: Live run — Edge X-ray (contrast) + 落盘
+
+**Files:**
+- (produces `report-cards/Edgedev-Edge-Computing-Engine-xray.md`; data feeds Task 8 example)
+
+- [ ] **Step 1: Run fair.py on Edge**
+
+```bash
+cd skills/gitlink-research-fair/scripts
+PYTHONUTF8=1 PYTHONIOENCODING=utf-8 python fair.py --owner Edgedev --repo Edge-Computing-Engine > _edge.json 2>&1
+PYTHONUTF8=1 python -c "
+import json
+d=json.load(open('_edge.json',encoding='utf-8'))
+print('Edge | paper:', d['paper'].get('arxiv_id'), '| methods:', d['methods'][:3], '| frameworks:', d['frameworks'][:2])
+print('license file:', d['license']['file'], '| files:', d['files_count'])
+"
+```
+Expected: paper arxiv likely None (Edge has no arxiv), methods incl autodiff/CNN-ish, license file True. Keep `_edge.json`.
+
+- [ ] **Step 2: LLM-render Edge X-ray + save**
+
+Render the Edge X-ray (controller). The **license-conflict finding** (README "禁止闭源商用" vs Apache LICENSE — note: fair.py detects `license.file=True`; the conflict is read from README content during rendering) must appear as a key finding. Verdicts will differ from Feature_Critic (e.g., 论文溯源 ❌, no paper). Save to `report-cards/Edgedev-Edge-Computing-Engine-xray.md`. The two reports must look visibly different (proves not 千篇一律).
+
+- [ ] **Step 3: No commit** — proceed to Task 8.
+
+---
+
+## Task 8: examples (2 docs) + remove v1 example
+
+**Files:**
+- Create: `skills/gitlink-research-fair/examples/feature-critic-xray.md`
+- Create: `skills/gitlink-research-fair/examples/edge-xray.md`
+- Remove: `skills/gitlink-research-fair/examples/research-fair-workflow.md` (v1 songhui18)
+
+- [ ] **Step 1: Write feature-critic-xray.md (using Task 6 outputs)**
+
+Real walkthrough: the fair.py command + the画像摘要 + the full X-ray report (from `report-cards/liyiying10-Feature_Critic-xray.md`) + a "答辩演示脚本" section (what to say when demoing live). No placeholders — all real values (arxiv 1901.11448, PACS/Visual Decathlon, etc.).
+
+- [ ] **Step 2: Write edge-xray.md (using Task 7 outputs)**
+
+Real walkthrough: fair.py command + 画像 + X-ray report (license-conflict finding). Emphasize the contrast with Feature_Critic (different profile → different report).
+
+- [ ] **Step 3: Remove v1 example + clean temp + commit**
+
+```bash
+cd "C:\Users\CWQ98\Desktop\演化与运维\gitlink-cli"
+git rm skills/gitlink-research-fair/examples/research-fair-workflow.md
+rm -f skills/gitlink-research-fair/scripts/_fc.json skills/gitlink-research-fair/scripts/_edge.json
+rm -rf skills/gitlink-research-fair/scripts/__pycache__
+git add skills/gitlink-research-fair/examples/feature-critic-xray.md skills/gitlink-research-fair/examples/edge-xray.md
+git commit -m "feat(fair-v2): examples Feature_Critic+Edge 真实X光走查；移除 v1 songhui18 示例"
+```
+
+---
+
+## Task 9: README + workflow update + register + acceptance + push
+
+**Files:**
+- Modify: `skills/README.md`
+- Modify: `skills/gitlink-workflow/SKILL.md`
+
+- [ ] **Step 1: Update README fair row**
+
+In `skills/README.md`, change the gitlink-research-fair row to:
+```markdown
+| **gitlink-research-fair** | 科研软件 X 光 | fair.py 抽论文/数据/复现/引用画像，四维裁决 + 真科研图谱 + 特有关键发现报告，可选处方 PR |
+```
+
+- [ ] **Step 2: Update workflow link text**
+
+In `skills/gitlink-workflow/SKILL.md`, change "科研软件 FAIR 体检" → "科研软件 X 光（fair v2）".
+
+- [ ] **Step 3: Re-register + acceptance (spec §16)**
+
+```bash
+bash scripts/setup-skills.sh   # re-link (no-op if already linked)
+ls ~/.claude/skills/ | grep gitlink-research-fair
+```
+Verify spec §16:
+- [ ] SKILL.md（重写）+ fair.py + test_fair.py + REFERENCE.md（重写）+ 2 examples 齐全
+- [ ] `python fair.py --owner liyiying10 --repo Feature_Critic` 出合法画像 JSON（arxiv 1901.11448）
+- [ ] test_fair.py 6 测试全过
+- [ ] Feature_Critic X 光：论文✅/数据⚠️/复现⚠️/引用❌ + 真图谱 ≥6 节点 + 关键发现 ≥3
+- [ ] Edge X 光：license 冲突作为发现；与 Feature_Critic 报告明显不同
+- [ ] 报告落盘 report-cards/
+- [ ] v1 五轴雷达/假 KG 已从 SKILL.md 移除（grep 0）
+- [ ] README/workflow 描述更新
+
+- [ ] **Step 4: Commit + push (updates PR #5)**
+
+```bash
+git add skills/README.md skills/gitlink-workflow/SKILL.md
+git commit -m "docs(fair-v2): README/workflow 更新为「科研软件 X 光」+ 注册"
+git push myfork feat/gitlink-research-fair
+```

docs/superpowers/specs/2026-06-18-intelligent-code-review-design.md

@@ -0,0 +1,216 @@
+# 智能代码审查 Skill（gitlink-review）设计
+
+- **日期**：2026-06-18
+- **状态**：已批准，待编写实现计划
+- **作者**：CWQ + Claude
+- **定位**：子任务二核心场景之一——智能代码审查；纯 Markdown skill（SKILL.md + REFERENCE.md + examples/），无需 Go 代码
+
+---
+
+## 1. 背景与目标
+
+gitlink-cli 现有 16 个 skill 中**没有专门的代码审查 skill**：`gitlink-workflow` 的"PR 全流程"仅在 Step 5 浅层提及（`pr +files` + "给出审查意见"）。README 列的智能 skill 只有 health / changelog / triage 三件套。
+
+本 skill 填补该缺口：引导 AI agent 对指定 GitLink PR 执行**结构化、多视角、低误报**的代码审查，并把结构化审查意见作为评论发布到 PR。
+
+**实用性目标**：信噪比高、误报少、每条发现可执行（带 `file:line` + 修复建议 + 理由）、能稳定落地到 GitLink（`pr +comment` 已验证可用）。
+
+**创意性目标**：多视角"评审团" + 对抗式自检（自带质量门，直击 AI code review 的误报老大难）+ 生产风险（oncall）视角。
+
+## 2. 非目标（YAGNI）
+
+- 不做 CI / webhook 自动触发——按需由 agent 调用
+- **绝不自动 approve / merge**
+- 不做跨仓库批量（单 PR 为主；批量留作可选入口，不在首版实现）
+- 不修改 gitlink-cli 的 Go 代码——纯 skill 交付物
+
+## 3. 关键决策（用户确认）
+
+| 决策点 | 选择 |
+|--------|------|
+| 评论形式 | 总结评论为主（`pr +comment`，可靠）+ 行内评论增强（reviews API 可用时附加，否则降级） |
+| 发表策略 | 默认预览、确认后发；`--auto` 跳过确认（仍受置信度门控） |
+| 审查引擎 | 多视角评审团（6 视角全量）+ 对抗式自检 |
+
+## 4. 文件清单与改动范围
+
+| 文件 | 动作 | 内容 |
+|------|------|------|
+| `skills/gitlink-review/SKILL.md` | 新增 | 主管道、6 视角、对抗自检、降噪、输出格式、安全护栏、命令接口 |
+| `skills/gitlink-review/REFERENCE.md` | 新增 | 各视角检查清单、严重度评级表、字段映射、reviews API 实测结论与降级、噪音文件规则 |
+| `skills/gitlink-review/examples/review-workflow.md` | 新增 | 真实 PR 走查（含植入缺陷验证、真实输出） |
+| `skills/README.md` | 修改 | 智能技能表新增 gitlink-review |
+| `skills/gitlink-workflow/SKILL.md` | 修改 | 浅层 Code Review 步骤链回本 skill |
+
+## 5. 命令接口
+
+skill 由 agent 调用，参数（约定，非 CLI 子命令）：
+
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--owner/--repo` | 自动从 cwd 解析 | 目标仓库 |
+| `--id` | 必填 | PR 编号（`pull_request_number`，网页 URL `/pulls/N`） |
+| `--lenses` | 全 6 视角 | 指定子集，如 `correctness,security` |
+| `--auto` | 关 | 跳过预览确认，直接发布（仍受置信度门控） |
+| `--inline` | 关 | 尝试附行内评论（reviews API 可用时） |
+| `--max-findings` | 12 | 报告卡上限，防刷屏 |
+| `--refresh` | 关 | 即使检测到旧审查哨兵也重审并更新 |
+
+## 6. 数据流（8 步管道）
+
+```
+① 取上下文   pr +view / pr +files / pr +diff（核心输入）/ repo +info（语言校准、约定）
+② 降噪过滤   跳过生成代码 / vendor / lock / 纯重命名 / 二进制 / minified
+③ 多视角分析 6 视角并行扫 diff → 候选发现
+④ 对抗式自检 逐条尝试反驳 → 丢弃不成立/低置信
+⑤ 综合      跨视角去重、按 严重度×likelihood 排序、封顶 → 报告卡
+⑥ 预览      默认展示给用户，确认后发（--auto 跳过）
+⑦ 发布      pr +comment 发总结评论（可靠）；--inline 且 reviews API 可用 → 附行内评论，否则降级
+⑧ 幂等      评论内嵌哨兵 <!-- gitlink-review v1 -->，重跑检测旧审查 → 提议更新而非刷屏
+```
+
+每条候选发现的数据结构（内部）：
+
+```json
+{
+  "lens": "correctness",
+  "severity": "high",          // high(🔴) | medium(🟡) | low(🔵)
+  "likelihood": "likely",      // likely | possible | unlikely
+  "file": "auth/login.go",
+  "line": 42,
+  "what": "空指针未判",
+  "why": "GetUser 在 token 无效时返回 nil，此处直接解引用",
+  "fix": "if u := GetUser(t); u != nil { ... }",
+  "confidence": "high"         // high | medium | low
+}
+```
+
+## 7. 评审团 6 视角
+
+| 视角 | 盯什么 |
+|------|--------|
+| 🔴 正确性 (correctness) | 逻辑错、边界、空值、并发竞态、资源泄漏、错误处理、类型转换 |
+| 🔒 安全 (security) | 注入（SQL/命令/XSS）、鉴权越权、密钥/Token 泄露、路径穿越、不安全反序列化、弱加密 |
+| ⚡ 性能 (performance) | N+1 查询、无谓拷贝、O(n²)/嵌套循环、热路径分配、缺索引、大对象常驻 |
+| 🧪 测试 (tests) | 新增/改动代码有无对应测试、边界用例、断言是否有效、mock 是否合理 |
+| 🧹 可维护性 (maintainability) | 命名、重复代码、圈复杂度、抽象边界、与既有约定一致性 |
+| 🚨 生产风险 (prod-risk) | "合并后凌晨3点哪会炸"——可观测性/日志、回滚能力、破坏性变更（API/DB/schema）、配置依赖、降级路径 |
+
+各视角的展开检查清单写入 `REFERENCE.md`。
+
+## 8. 对抗式自检（质量门，创意核心）
+
+对每条候选发现，agent 自我问以下问题，任一成立即丢弃或降级：
+
+1. **语境已处理**：完整上下文里是否已有保护（外层判空、上游校验、框架机制）？
+2. **风格冒充 bug**：这是真缺陷还是个人风格偏好？若是偏好，降级为 🔵 nit 或丢弃。
+3. **幻觉检查**：我引用的 API/函数签名/语言行为是否真实存在？（不确定则不报，或标注"待确认"）
+4. **重复/已被覆盖**：是否与另一视角的发现其实是同一问题？
+
+**置信度门控**：`confidence=low` 且 `severity≠high` → 丢弃。这保证只发高信号发现。
+
+## 9. 降噪规则（自动跳过，不审查）
+
+- 生成代码：`*.gen.go`、`*.pb.go`、`*_generated.*`、`*.min.js`、dist/、build/、target/
+- 第三方：`vendor/`、`third_party/`、`node_modules/`
+- 锁文件：`go.sum`、`package-lock.json`、`yarn.lock`、`pnpm-lock.yaml`、`Cargo.lock`
+- 纯重命名/移动（无内容变更）
+- 二进制文件、图片、字体
+
+跳过时在报告卡"范围"行注明跳过数量，不静默吞掉。
+
+## 10. 输出格式（报告卡，作为 PR 评论发布）
+
+```markdown
+🤖 **gitlink-review 报告**
+
+**结论**：⚠️ 建议修改（2 🔴阻塞 / 4 🟡建议 / 3 🔵nit）
+**范围**：5 文件，+120/-30（跳过 2 噪音文件）｜视角：6｜自检丢弃：3
+
+| 严重度 | 视角 | 位置 | 问题 |
+|:---:|---|---|---|
+| 🔴 | 正确性 | auth/login.go:42 | 空指针未判 |
+| 🔴 | 安全 | config.go:8 | 硬编码 Token |
+| 🟡 | 性能 | list.go:88 | 循环内重复查询 |
+
+### 🔴 阻塞（合并前需处理）
+1. **[正确性] auth/login.go:42** — `GetUser(token)` 在 token 无效时返回 nil，此处直接解引用会 panic。
+   - **建议**：`if u := GetUser(t); u != nil { ... }`
+2. **[安全] config.go:8** — 硬编码 Token，存在泄露风险。
+   - **建议**：改从环境变量读取 `os.Getenv("GITLINK_TOKEN")`
+
+### 🟡 建议
+…
+
+### 🔵 nit
+…
+
+### ✅ 未发现问题的方面
+- 安全：未发现注入点
+- 生产风险：无破坏性 API 变更
+
+---
+<!-- gitlink-review v1 | pr:5 | lenses:6 | refuted:3 | sha:<head-sha> -->
+*由 gitlink-review skill 生成；本评论为预览确认后发布。*
+```
+
+哨兵 `<!-- gitlink-review v1 | pr:N | lenses | refuted | sha -->` 用于幂等与"针对哪个 commit"标识。
+
+## 11. 安全护栏
+
+- 默认预览确认；`--auto` 跳过确认但**仍受置信度门控**
+- **绝不自动 approve / merge**（即使 `--auto`）
+- 行内评论仅在显式 `--inline` 且 reviews API 实测可用时发，否则不发
+- `--max-findings` 封顶，防刷屏
+- 所有写操作需认证（遵循 `gitlink-shared`）
+- 跳过自己作者本人的 PR 时可提示（避免自审盲区），但不强制阻断
+
+## 12. 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| `pr +diff` 无差异 | 报告"无可审查变更"，不发评论 |
+| PR 不存在/无权限 | 清晰错误信息，不发评论 |
+| reviews API 404（已知 `api POST` bug） | 记录、跳过行内、总结评论照发 |
+| diff 过大（> 阈值） | 抽样审查 + 标注"部分审查（仅 X/Y 文件）" |
+| `pr +comment` 发布失败 | 把报告卡原文输出给用户手动粘贴 |
+| 置信度全部 low | 报告"未发现高置信问题"，列出待人工确认项 |
+
+## 13. 幂等与去重
+
+- 评论内嵌哨兵；重跑时先 `pr +view`/取评论检测哨兵
+- 已存在旧审查：默认提议"更新"（`--refresh` 才覆盖），避免重复发
+- `sha` 字段记录审查所基于的 head commit；PR 有新提交时提示"审查已过期，建议重审"
+
+## 14. 验证计划（真实数据，沿用其他 skill 惯例）
+
+建一个**植入已知缺陷的小测试 PR**，包含：
+
+1. 一处空指针/越界（验证 🔴 正确性视角命中）
+2. 一处硬编码密钥（验证 🔒 安全视角命中）
+3. 一处缺测试的新函数（验证 🧪 测试视角命中）
+4. 一处"看起来像 bug 但语境已处理"的伪阳性（验证对抗自检丢弃）
+
+验证项：
+- [ ] 每个植入缺陷被对应视角命中
+- [ ] 伪阳性被对抗自检丢弃（refuted 计数 +1）
+- [ ] 哨兵正确内嵌，重跑不刷屏
+- [ ] 总结评论通过 `pr +comment` 成功发布
+- [ ] reviews API 实测：记录可用/不可用结论写入 REFERENCE.md
+- [ ] `--auto` 与预览两种模式均验证
+
+结果（真实输出）写入 `examples/review-workflow.md`。
+
+## 15. 风险与未决
+
+- **reviews API 可用性**：`gitlink-triage` 记录 `api POST` 有 URL 注入 bug 导致 404。本 skill 的行内评论依赖 `POST /:owner/:repo/pulls/:id/reviews`，实现时**必须实测**；不可用则设计已内建降级（仅总结评论）。结论写入 REFERENCE.md。
+- **token 消耗**：6 视角全量较重；通过 `--lenses` 子集和 `--max-findings` 控制。
+- **行号偏移**：diff 行号 vs 文件行号的映射需在 REFERENCE.md 给出规则，保证 `file:line` 准确。
+
+## 16. 验收标准
+
+1. `skills/gitlink-review/` 三件套齐全，结构与 triage/health 等智能 skill 一致
+2. 测试 PR 的 4 个植入点验证全部通过（3 命中 + 1 丢弃）
+3. 总结评论在真实 PR 成功发布，含哨兵
+4. reviews API 可用性有明确实测结论
+5. README 与 workflow 链接更新

docs/superpowers/specs/2026-07-01-gitlink-research-fair-design.md

@@ -0,0 +1,249 @@
+# 科研软件 FAIR 体检 Skill（gitlink-research-fair）设计
+
+- **日期**：2026-07-01
+- **状态**：已批准，待编写实现计划
+- **作者**：CWQ + Claude
+- **定位**：子任务四（应用 GitLink 辅助科研，加分项）——把 GitLink 科研仓库转化为"FAIR 体检报告卡 + 自动修复 + 可复现证书"；纯 Markdown skill（SKILL.md + REFERENCE.md + examples/），无需 Go 代码
+- **架构中心**：报告卡中心（FAIR 体检报告为 hero 产出，KG/处方/证书/OpenAlex 为其栏目）
+
+---
+
+## 1. 背景与目标
+
+子任务四要求把 gitlink-cli 与学术科研结合，做一个"科研分析层"skill。gitlink-cli 现有 17 个 skill 都是**数据采集/运维型**（issue/pr/search/user/pm/review/health/license/changelog/ci/repo/branch/release/triage/workflow/org/shared），缺少把仓库数据**升维成科研洞察**的能力。
+
+本 skill 填补该缺口：对任意 GitLink 科研仓库执行**结构化、有学术标准依据、低误报**的"FAIR + 可复现性"诊断，产出一张**体检报告卡**，并对发现的缺口**自动开 PR 修复**、用 **SWH/commit 锚定版本**签发**可复现证书**。
+
+**实用性目标**：评分有据（对标 FAIR4RS + howfairis）、每项检查可追溯证据（哪个 gitlink-cli 字段/文件）、缺口可执行修复（生成 CITATION.cff/codemeta.json/LICENSE/复现章节）、能在真实 GitLink 科研仓库跑通。
+
+**创意性目标**：四幕剧本（**诊断 → 处方 → 证书**，v1 不含排行）+ 中国代码托管平台**原生**的科研软件诊断（现有 howfairis/Codefair 只跑 GitHub）+ 知识图谱 + OpenAlex 溯源。
+
+## 2. 非目标（YAGNI）
+
+- **不做多仓库 FAIR 排行榜**（第四幕"排行"留 v2）
+- 不做 CI/webhook 自动触发——按需由 agent 调用
+- **绝不自动 merge** 修复 PR；处方默认预览确认，`--auto` 仅跳过预览
+- **绝不 force-push、绝不碰原仓库**——处方只对 fork 开 PR
+- 不修改 gitlink-cli 的 Go 代码——纯 skill 交付物
+- 不重复 `gitlink-license` 已有能力——许可证/安全子项**委托复用**它
+
+## 3. 关键决策（用户确认）
+
+| 决策点 | 选择 | 理由 |
+|--------|------|------|
+| v1 范围 | 诊断 + 处方 + 证书（不做排行） | 保留最惊艳的"自动修复闭环"，避开多仓库验证成本 |
+| 验证仓库 | `songhui18/ICCV2021论文复现`（处方走 fork） | "论文复现"主题与"可复现性"内核共振；19⭐真实活跃；无 license=现成缺口 |
+| 架构中心 | 报告卡中心（A 方案） | 最强"信服"载体（FAIR4RS 逐项可核对）；避免评委认知过载 |
+| 许可证/安全支柱 | 复用 `gitlink-license` | 已有 4 维评分+模式库，不重复造 |
+| 发表策略 | 默认预览、确认后开 PR；`--auto` 跳过预览 | 沿用既有"先预览再执行"偏好 |
+
+## 4. 学术依据与差异化（"信服"骨架，每条对抗式核验过）
+
+| 支撑点 | 文献（已核验） | 在 skill 中的用途 |
+|--------|----------------|------------------|
+| 科研软件 FAIR 原则有国际标准，F1/F1.2/F2/R1.1/R1.2/R2 可逐条机器校验 | Barker et al. 2022, *Nature Sci Data* [s41597-022-01710-x]（3-0 ✓×3） | 评分骨架 |
+| 可复现性危机铁证：2024 ICLR/ICML/NeurIPS 论文仅 19.5% 提供官方代码 | PaperCoder [arXiv:2504.17192] Table 9（2-0 ✓） | 开场"为什么重要"硬数据 |
+| 结构化对比表/报告卡是有学术先例的成熟范式 | ORKG (K-CAP 2019)（3-0 ✓） | 报告卡范式出处 |
+| 仓库数据→RDF 知识图谱的实体/关系 schema 可借鉴（13 实体/47 关系/794 万三元组） | LPWC, ISWC 2023（3-0 ✓） | KG 本体设计 |
+| 版本锚定用 SWH-ID（Merkle DAG + git 兼容哈希，本地可生成） | Di Cosmo et al. ICMS 2020 [PMC7340894]（3-0 ✓） | 证书版本锚定 |
+| 作者/机构溯源直接调 OpenAlex REST 反查（无需自建索引） | Priem et al. 2022 [arXiv:2205.01833]（3-0 ✓） | 溯源富集 |
+| 仓库级 KG 问答在真实仓库跑通（四实体+SZZ+Cypher，CoT 50%→90%） | Repo-KG [arXiv:2412.03815]（3-0 ✓） | KG+推理可行性基线 |
+| 科研软件"可用+可复现"有现成徽章体系可对标 | ACM Artifact Badging [acm.org]（5 核验） | 证书/徽章锚 |
+| 直接竞品 howfairis（5 维：repository/license/registry/citation/checklist）只跑 GitHub | fair-software/howfairis | 差异化：GitLink 原生 |
+
+**差异化叙事（必须咬死）**：*第一个原生跑在中国代码托管平台上、面向任意科研仓库的实时 FAIR+可复现性诊断 → 自动修复 → 证书闭环，纯 Agent Skill 形态。* 与 Software Heritage（重存档）、Papers With Code（重列表）、OpenAlex（重文献）、ASCL（重天体软件）正交。
+
+**战略时机**：据 TIB 2025-10 博客 *Papers With Code went offline*，PWC 出现下线/不稳定，社区在找替代——中国平台原生科研软件诊断工具是真空（⚠️ 单一二手源，pitch 前自行复核 PWC 状态）。
+
+**诚实边界（不可过度宣称）**：
+1. FAIR4RS 原文自称 **aspirational**，F1.2/R1.1/R1.2/R2 可逐条校验，但"全自动合规验证器"社区仍在建（SOMEF/FAIR EVA）——本 skill 是**自建 checker**，这本身是贡献，不说"套用现成标准工具"。
+2. **FAIRness ≠ 可复现性**（FAIR 是必要非充分）——报告卡里 Repro **单独成轴**，不混为一谈。
+3. OpenAlex 自 2025-02 起 freemium（免费档 ~$1/day、需 API key、100 req/s）——批量控量，单仓演示够。
+4. **两条已被证伪的论点禁止使用**：①"MSR 六分类法"、②"FAIR 分高→被引更多"的因果（研究已剔除）。
+5. 文献多跑在 GitHub——"在 GitLink 上复刻"是合理外推，必须真机跑通闭环。
+
+## 5. 文件清单与改动范围
+
+| 文件 | 动作 | 内容 |
+|------|------|------|
+| `skills/gitlink-research-fair/SKILL.md` | 新增 | 主管道、5 轴 rubric、报告卡模板、处方护栏、证书、KG/OpenAlex、错误降级、命令接口 |
+| `skills/gitlink-research-fair/REFERENCE.md` | 新增 | FAIR4RS 逐条映射表、评分算法、KG schema(triples)、OpenAlex 字段、SWH-ID 说明、**学术引用库**、诚实边界 |
+| `skills/gitlink-research-fair/examples/research-fair-workflow.md` | 新增 | `songhui18/ICCV2021论文复现` 真实走查（报告卡+处方 PR+证书+截图） |
+| `skills/README.md` | 修改 | 新增 gitlink-research-fair（科研辅助类） |
+| `skills/gitlink-workflow/SKILL.md` | 修改（可选） | 在合适处链回本 skill |
+
+## 6. 命令接口
+
+skill 由 agent 调用，参数（约定，非 CLI 子命令）：
+
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--owner/--repo` | 自动从 cwd 解析 | 目标科研仓库 |
+| `--lenses` | 全 5 轴 | 子集，如 `F,A,Repro` |
+| `--auto` | 关 | 跳过预览直接开 PR（仍受护栏：不 force-push/不碰原仓/不 merge） |
+| `--no-fork` | 关 | 不开 PR，处方物料落本地文件供手动处理 |
+| `--enrich` | 开 | 启用 OpenAlex 溯源（限流则自动降级） |
+| `--refresh` | 关 | 即使存在旧报告哨兵也重评 |
+
+## 7. 数据流（8 步管道）
+
+```
+① 取上下文   repo +info（元数据/license/topics/has_dataset）· file +list（关键文件清单）
+② 深采       file +get 读 LICENSE/CITATION.cff/codemeta.json/README/requirements/Dockerfile ·
+             release +list（版本）· commit +list（SHA）· issue/pr +list（协作）· member +list（作者）
+③ FAIR 评分  M1 逐项评 5 轴（许可证/安全子项委托 gitlink-license 流程）→ ✓/✗/⚠/⊥ + 证据 + 置信度
+④ KG 构建    M2 四实体(Repo/Contributor/File/Commit)+Paper/Dataset/License 节点 → Mermaid 小图 + triples
+⑤ 溯源富集   M3 从 README/CITATION 抽论文 → OpenAlex REST 反查作者/机构（best-effort）
+⑥ 渲染报告卡 hero：等级+雷达图+5 轴逐项表+处方摘要+证书栏+KG 小图+溯源栏
+⑦ 处方       M4 对 ✗/⚠ 项生成 CITATION.cff/codemeta.json/LICENSE/复现章节 → 默认预览
+⑧ 发布       确认后对 fork 开 PR（PR body 带报告卡摘要）+ 出 SWH/commit 锚定证书；内嵌哨兵
+```
+
+每项检查的内部数据结构：
+
+```json
+{
+  "axis": "A", "id": "A2", "title": "有明确许可证",
+  "status": "✗",            // ✓满足 | ✗缺失 | ⚠部分 | ⊥无字段(跳过)
+  "evidence": "repo +info.license_id 为空；根目录无 LICENSE 文件",
+  "confidence": "high",      // high | medium | low
+  "fix": "建议添加 MIT 或木兰 PSL v2 LICENSE（处方可自动生成）"
+}
+```
+
+## 8. 评分 Rubric（核心）
+
+> 对标 **FAIR4RS（Nature Sci Data 2022）+ howfairis 5 维**。诚实声明：这是**适配版**评分（社区尚无认证级自动校验器），不是官方认证。**FAIR ≠ 可复现**，故 Repro 单独成轴。
+
+每项判定 `✓满足 / ✗缺失 / ⚠部分 / ⊥GitLink 无该字段(跳过并标注)`，附**证据来源** + 置信度。
+
+| 轴 | FAIR4RS 映射 | 检查项（证据） |
+|----|--------------|----------------|
+| **F 可发现** | F1/F2 | F1 清晰描述(repo desc 非空且≥20 字) · F2 话题/关键词(topics) · F3 标准元数据(codemeta.json/.zenodo.json) |
+| **A 可访问** | F1.2/R1.1 | A1 公开(is_public) · A2 有许可证(license_id/LICENSE)[gitlink-license] · A3 有版本发布(release tag) · A4 持久标识(commit SHA/SWH-ID) |
+| **I 可互操作** | R1.1/I | I1 依赖清单(requirements/go.mod/package.json/environment.yml) · I2 许可证机器可读(SPDX/license_id) · I3 元数据标准格式(CITATION.cff/codemeta) |
+| **R 可复用** | R1.2/R2 | R1 溯源/引用(CITATION.cff/README 引用) · R2 README 质量(安装+使用) · R3 源码声明[gitlink-license] · R4 无敏感泄露[gitlink-license] |
+| **Repro 可复现** ⭐差异轴 | （独立于 FAIR） | Rep1 环境锁(锁版本/Dockerfile) · Rep2 数据集说明(has_dataset/README 引用) · Rep3 复现步骤(README 章节) · Rep4 版本固定 · Rep5 入口可执行(main/Makefile/CLI) |
+
+**评分算法**：轴分 = 满足项 / (总项 − ⊥项)（⚠ 计 0.5）；总评 = 5 轴均分 → 等级 **A🟢≥80 / B🟡60-79 / C🔴<60**。若某轴全部为 ⊥（GitLink 无任何可用字段），该轴记 N/A 并从总评均分中排除。`⊥` 项在报告卡"范围"行注明"GitLink 无 X 字段，已跳过 N 项"，不静默吞掉。`confidence=low` 的项标注"待人工确认"，不进总评分计算。
+
+## 9. 报告卡格式（hero 产出）
+
+```markdown
+🏥 **gitlink-research-fair 科研软件体检报告**
+
+**总体**：B 🟡 64/100（F 70 · A 50 · I 40 · R 80 · Repro 80）｜版本锚定：commit a1b2c3d（无 release）
+**范围**：检查 19 项（✓11 · ✗4 · ⚠2 · ⊥2，跳过：GitLink 无 codemeta 字段）
+
+### 五轴雷达
+        Repro 80
+      /          \
+  R 80            F 70
+      \          /
+   I 40 —— A 50
+
+### 逐项（节选）
+| 轴 | 项 | 状态 | 证据 |
+|---|---|:---:|---|
+| A | A2 有许可证 | ✗ | license_id 空，根目录无 LICENSE |
+| Repro | Rep3 复现步骤 | ⚠ | README 有"使用"但无完整复现流程 |
+| F | F1 清晰描述 | ✓ | repo desc 详述 CV 顶会论文复现合集 |
+
+### 🔧 处方（可自动修复 3 项）
+- [R1] 生成 CITATION.cff（从 README/作者推断）
+- [A2] 添加 LICENSE（MIT / 木兰 PSL v2 任选）
+- [Rep3] 补"复现"章节（从入口推断）
+
+### 📜 可复现证书
+锚定版本：git+commit a1b2c3d ｜ FAIR: B 🟡 ｜ Repro: ⚠ 部分可复现
+（完整 SWH-ID 需 swh-identify；本次用 commit-SHA 锚定）
+
+### 🔗 溯源（OpenAlex）
+未找到关联论文（README 未含可识别 DOI/标题）— 已跳过
+
+### 🧬 科研关系图（KG）
+[Mermaid: Repo—contributes→Contributor×3 · Repo—licensed?—? · Repo—has→Dataset? ]
+
+---
+<!-- gitlink-research-fair v1 | repo:songhui18/ICCV2021论文复现 | grade:B | sha:a1b2c3d -->
+*由 gitlink-research-fair skill 生成。*
+```
+
+哨兵 `<!-- gitlink-research-fair v1 | repo | grade | sha -->` 用于幂等与"基于哪个 commit"标识。
+
+## 10. 处方模块（M4，惊艳闭环）与安全护栏
+
+对可自动修复的 ✗/⚠ 项生成物料：
+
+| 缺口 | 生成物 |
+|------|--------|
+| F3 缺 codemeta.json | 从 repo +info + README 生成 codemeta.json 草稿 |
+| R1 缺 CITATION.cff | 从 README/作者生成 CITATION.cff 草稿 |
+| A2 缺 license | 给 MIT / 木兰 PSL v2 模板（任选） |
+| Rep1 缺依赖锁 | 从 import 扫描建议 requirements.txt |
+| Rep3 缺复现章节 | 从入口/README 生成"复现"章节草稿 |
+
+**安全护栏**（沿用既有偏好）：
+- 默认全量预览 → 用户确认 → 对 **fork**（`caoweiqiong/ICCV2021论文复现`）开一个 PR，PR body 带报告卡摘要
+- `--auto` 跳过预览，但**永不 force-push、永不碰原仓库、永不自动 merge**
+- `--no-fork`：不开 PR，物料落本地文件供手动处理
+- 所有写操作需认证（遵循 `gitlink-shared`）
+
+## 11. 证书（SWH/commit 锚定）
+
+- 版本锚定：HEAD commit SHA + release tag（若有）
+- 若环境有 Python `swh.model` → 算完整 SWH-ID；否则用 `git+<commit-SHA>` 锚定，报告卡标注"完整 SWH-ID 需 swh-identify"
+- 证书内容：仓库、锚定版本、FAIR 等级、Repro 状态、生成时间、可引用条目（对标 ACM Artifact Badge 的 Available/Functional 风格）
+- 嵌入报告卡"证书栏"；亦可单独输出可复制引用文本
+
+## 12. KG + OpenAlex（支撑栏，非主角）
+
+- **KG**：四实体（Repo / Contributor / File / Commit）+ Paper / Dataset / License 节点；关系 contributes-to / authored / depends-on / cites / licensed-under / version-at。输出 Mermaid 小图（报告卡内嵌）+ triples JSON（REFERENCE 给 schema）。借鉴 LPWC 实体/关系划分与 Repo-KG 四实体。
+- **OpenAlex**：从 README/CITATION 抽论文标题/作者/DOI → REST `/works` 反查作者机构/发表载体/概念。降级：freemium 限流或抽不到 → 跳过，标注"未找到关联论文"，不影响主报告卡。
+
+## 13. 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| `file +tree --recursive` 返回空 | 手动递归子目录（gitlink-license 已知坑） |
+| 无 LICENSE | A2 ✗，处方建议模板 |
+| OpenAlex 限流/无果 | 跳过溯源栏标注 |
+| `swh.model` 不可用 | commit-SHA 锚定 + 标注 |
+| 仓库过大 | 抽样文件 + "部分审查"标注 |
+| fork 失败/无写权限 | 处方物料落本地文件供手动 PR |
+| 二进制是 npm 旧版 | 强制 `./gitlink-cli` 或提示 `go build` 重建 |
+| 仓库私有/无权限 | 清晰错误，不发报告 |
+| 报告/PR 发布失败 | 报告卡原文交用户手动粘贴 |
+
+## 14. 验证计划（真实数据）
+
+主目标：**`songhui18/ICCV2021论文复现`**（论文复现合集，19⭐/17fork，Python/CUDA/Jupyter，无 license）。
+
+验证项：
+- [ ] 报告卡正确给出 B/C 级（因无 license，A2 ✗），5 轴评分有区分度
+- [ ] 处方：fork 后成功开 1 个 PR（CITATION.cff + LICENSE + 复现章节），PR body 含报告卡摘要
+- [ ] 证书：commit-SHA 锚定正确（若有 swh.model 则给完整 SWH-ID）
+- [ ] **对照**：跑一个有 license 的仓库（`leejt/图神经网络快速开发测试工具` 或 `Edgedev/Edge深度学习框架`）→ 预期 A 级，证明评分有区分度
+- [ ] **边界**：私有仓 / 空仓 / 无 README → 降级正确，不崩溃
+- [ ] OpenAlex 溯源：能抽到论文则富集，抽不到则标注降级
+- [ ] KG：Mermaid 图 + triples JSON 正确生成
+
+结果（真实输出 + 截图）写入 `examples/research-fair-workflow.md`。
+
+## 15. 风险与未决
+
+- **GitLink 元数据粒度**：FAIR4RS 某些条目（如 R2 限定引用、标准 codemeta）GitLink 无专用字段 → 用 ⊥ 跳过 + 文件探测 + LLM 推断兜底，明确标注"推断 vs 已验证"。实现时需在真实仓库上确认哪些项可机器读取。
+- **OpenAlex freemium 限额**：2025-02 起；单仓演示够，批量需控量。
+- **SWH-ID 完整性**：纯 Markdown+CLI 环境无 `swh.model` 时只能 commit-SHA 锚定；实现时探测依赖，降级已内建。
+- **GitLink GraphQL/REST 粒度**：Repo-KG 的 Cypher QA 需图数据库，本 skill **不引入图数据库**（纯 Mermaid+triples），KG 仅作可视化/溯源，不做复杂问答（YAGNI）。
+- **处方 PR 真实性**：对 fork 开 PR 需 `gitlink-cli repo +fork` + 本地改文件 + push + `pr +create`；需实测 fork→PR 链路是否顺畅，不通则降级 `--no-fork`。
+
+## 16. 验收标准
+
+1. `skills/gitlink-research-fair/` 三件套齐全，结构与 health/changelog/triage/review 等智能 skill 一致
+2. 5 轴 rubric 每项有明确证据来源 + 状态判定 + 兜底规则
+3. `songhui18/ICCV2021论文复现` 真实跑通：报告卡 + 处方 PR（对 fork）+ 证书，有截图
+4. 对照仓库评分有区分度（有 license → A，无 → B/C）
+5. 边界情况（私有/空/无 README）降级正确
+6. REFERENCE.md 含完整学术引用库 + 诚实边界（含 2 条禁用论点）
+7. README 登记 gitlink-research-fair

docs/superpowers/specs/2026-07-01-gitlink-spark-design.md

@@ -0,0 +1,212 @@
+# 文献-代码语义缺口挖掘机 Skill（gitlink-spark）设计
+
+- **日期**：2026-07-01
+- **状态**：已批准，待编写实现计划
+- **作者**：CWQ + Claude
+- **定位**：子任务四（应用 GitLink 辅助科研）的**第二部分**——与 `gitlink-research-fair`（评估已有）组成"科研辅助双联装"；本 skill 负责"**发现空白**"。
+- **形态**：skill + **可运行 python 脚本**（非纯 Markdown）；不写 Go。
+
+---
+
+## 1. 背景与目标
+
+`gitlink-research-fair` 解决"已有科研软件合不合格"。研究者还有个更核心的痛点：**找新的研究点**。他们读大量论文找灵感，却很难发现"**学术界的最新理论**"与"**开源社区的实际落地**"之间有哪些未被填补的空白——而这往往是低成本发论文/出成果的黄金地带。
+
+本 skill 填补该缺口：给一个研究领域，跨 **arXiv（学术）× GitLink（中文生态）× GitHub（全球）** 三源，挖出两类语义缺口，输出**空白学术机会报告**，并可一键 **fork+issue 起跑**。
+
+**实用性目标**（demo 级，非生产级全场景通用）：在 GNN 领域跑出 **2-3 个漂亮真实例子**供答辩展示；每条缺口带可查实证。
+
+**创意性目标**：从"paper→code gap"角度启发创新，区别于 Papers With Code（只链接、不挖缺口）/ OpenAlex（只文献）/ GitHub Archive（只存档）。
+
+## 2. 非目标（YAGNI）
+
+- **不做生产级全领域通用**——demo 级，GNN 跑通 2-3 例即可，不追求任意领域鲁棒
+- 不做实时订阅/webhook——按需 agent 调用
+- 不自动撰写论文/实现代码——只"提议机会 + 起跑 scaffold"
+- **绝不自动 merge**；起跑（fork+issue）默认预览，`--auto` 跳过预览
+- 不修改 gitlink-cli 的 Go 代码——skill + python 脚本
+- 不重复 `gitlink-research-fair`——fair 评估单仓 FAIR，spark 跨仓跨源挖缺口
+
+## 3. 关键决策（用户确认）
+
+| 决策点 | 选择 | 理由 |
+|--------|------|------|
+| v1 缺口类型 | **两种都做**（有理论无实现 + 有需求无解答） | 完整愿景；demo 级 2-3 例即可 |
+| 验证领域 | **图神经网络 GNN** | 三源最均衡：arXiv 海量、GitLink 少量（缺口真实）、GitHub 海量（对照鲜明） |
+| GitHub 全球对照 | **硬需求**，做 | 避免"GitLink 0 ≠ 全球空白"误导；含阈值过滤 |
+| 缺口匹配引擎 | **A. LLM 语义匹配为主** | 脚本抓真实数据，LLM 提缺口假设+附实证三件套；demo 级最轻最灵活 |
+| 起跑动作 | fork 最近基准 + issue 粘伪代码 todo | 对称 fair 处方 PR；强 demo 闭环 |
+| 交付方式 | 与 fair 同 PR #5（子任务四双联装） | 统一叙事 |
+
+## 4. 数据源与可行性（实测 2026-07-01）
+
+| 数据源 | 状态 | 用途 | 备注 |
+|--------|------|------|------|
+| arXiv API（HTTPS） | ✅ HTTP 200，返回论文条目 | ① 学术采：领域近 N 天论文 | 必须 HTTPS（HTTP 被沙箱阻断） |
+| `gitlink-cli search +repos` | ✅ | ② GitLink 仓库 | 用 identifier/关键词 |
+| `gitlink-cli issue +list`（逐仓库） | ✅ | ② GitLink open issue | 绕开 `search +issues`（返回 HTML 的坑） |
+| GitHub Search API | ✅ HTTP 200（total_count + items） | ③ 全球对照 | 未认证 10 req/min；用 `GITHUB_TOKEN` 提至 5000/h |
+| OpenAlex | ⚠ 间歇 503 | 引用计数（可选富集） | best-effort，降级跳过 |
+
+## 5. 文件清单
+
+| 文件 | 动作 | 内容 |
+|------|------|------|
+| `skills/gitlink-spark/SKILL.md` | 新增 | 4 阶段管道编排、缺口分类法、报告模板、GitHub 阈值规则、起跑护栏、错误降级、命令接口 |
+| `skills/gitlink-spark/scripts/spark.py` | 新增 | **可独立运行的数据融合脚本**：arXiv + gitlink-cli + GitHub API → 输出 JSON 给 LLM；含缓存与限流 |
+| `skills/gitlink-spark/REFERENCE.md` | 新增 | 缺口分类法细则、GitHub 阈值与分级、LLM prompt 模板、数据源实测结论、诚实边界 |
+| `skills/gitlink-spark/examples/spark-图神经网络.md` | 新增 | GNN 真实跑出的 2-3 缺口卡（含起跑截图） |
+| `skills/README.md` | 修改 | 智能技能表加 gitlink-spark |
+| `skills/gitlink-workflow/SKILL.md` | 修改（可选） | 专项 skill 链表加 gitlink-spark |
+
+## 6. 命令接口
+
+### spark.py（独立可运行）
+```bash
+python skills/gitlink-spark/scripts/spark.py \
+  --field "图神经网络" \
+  [--max-papers 10] \
+  [--gap-type both|theory|demand] \
+  [--github-token $GITHUB_TOKEN]   # 可选，未设则按 10/min 限流
+  # → stdout 输出融合 JSON：{papers, gitlink_repos, gitlink_issues, github_counts}
+```
+
+### skill 约定参数（SKILL.md，非 CLI flag）
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--field` | 必填 | 研究领域（如 `图神经网络`、`federated learning`） |
+| `--gap-type` | `both` | `theory` / `demand` / `both` |
+| `--max-papers` | 10 | arXiv 抓取论文上限（控制 GitHub 调用） |
+| `--auto` | 关 | 跳过预览直接起跑（仍受护栏） |
+| `--no-fork` | 关 | 只出报告，不起跑 |
+
+## 7. 数据流（4 阶段管道）
+
+```
+① 学术采   spark.py: arXiv HTTPS 抓领域近 90 天论文（标题/摘要/arxiv id/方法关键词）
+② GitLink  spark.py: gitlink-cli search +repos 抓领域仓库；
+                    对每个仓库 issue +list 抓 open issue（绕开 search+issues HTML）
+③ 全球对照 spark.py: GitHub Search API 对每个论文方法查 total_count + Top3 仓库
+④ 缺口匹配（LLM，SKILL.md 编排）
+            读 spark.py 输出的 JSON → 语义匹配两类缺口 → 每张带实证三件套 → 渲染机会报告
+⑤ 起跑（可选） 选定缺口 → repo +fork 基准 → issue +create 粘论文伪代码 todo
+```
+
+**职责切分**：`spark.py` 只抓**真实数据**（确定性、可复现）；**缺口发现**交给 LLM（语义判断），但必须附实证，受置信度门控。
+
+spark.py 输出 JSON schema：
+```json
+{
+  "field": "图神经网络",
+  "papers": [{"arxiv_id":"2403.xxxxx","title":"...","abstract":"...","published":"2024-03-15","method_keywords":[...]}],
+  "gitlink_repos": [{"owner":"leejt","repo":"GraphGallery","desc":"...","topics":[...]}],
+  "gitlink_issues": [{"repo":"leejt/GraphGallery","number":12,"subject":"...","status":"open","participants":3}],
+  "github_counts": [{"method":"graph attention XXX","total_count":2,"top":[{"full_name":"...","stars":3534}]}]
+}
+```
+
+## 8. 两类缺口分类法 + 实证三件套
+
+每张缺口卡**必须**带齐三件套，否则被置信度门控丢弃（防 LLM 编造）：
+
+### 类型 A：有理论无实现（paper → code gap）
+- **三件套**：① 论文 arxiv id + 标题 + 发表日期 ② GitLink 搜索查询串 + 命中数（0 或极少） ③ GitHub total_count + Top 仓库（参考基准）
+- **判定**：论文提出方法 M；GitLink 实现 0/极少；GitHub 按下面阈值分级。
+
+### 类型 B：有需求无解答（open issue → applied research gap）
+- **三件套**：① issue URL + 主题 + 讨论人数/状态 ② GitLink 无现成实现解此痛点 ③ GitHub 是否有成熟开源解（对照）
+- **判定**：领域仓库中存在"研究性痛点"open issue（排除"安装报错"类使用问题）；学术/GitHub 无现成解。
+- **降噪**：LLM 筛"研究性痛点"（含性能/可扩展性/新场景），排除使用/部署类。
+
+## 9. GitHub 全球对照与阈值（诚实核心，硬需求落地）
+
+防止"GitLink 0 ≠ 全球空白"误导。对每个"理论无实现"候选，按 GitHub total_count 分级：
+
+| GitHub total_count | 分级 | 报告行为 |
+|--------------------|------|----------|
+| `< 10` | **全球稀缺（真空白）** | 报为高价值缺口："GitLink 生态空白 × 全球稀缺 → 复现并开源到 GitLink，易成本平台标杆" |
+| `10–50` | **新兴（部分空白）** | 报为中等缺口："GitLink 空白，全球新兴（N 个），可做中文生态首个完整实现" |
+| `≥ 50` | **全球已成熟** | **不报为空白**，列入"✅ 已诚实排除"区："GitLink 虽 0，但 GitHub 已 N 个（含官方）→ 全球已成熟，非空白" |
+
+> 这个分级是本 skill 的诚实命门：宁可少报，不误报机会。GitHub 计数缓存（spark.py 按 method key 缓存去重）。
+
+## 10. 机会报告格式（hero）
+
+````markdown
+⚡ **gitlink-spark 机会报告：<field>**
+
+学术采：arXiv 近 90 天 N 篇 ｜ GitLink 仓库 M 个 ｜ GitHub 全球基线已对照
+生成时间：YYYY-MM-DD
+
+### 🧩 缺口 1 · 有理论无实现 [全球稀缺·高价值]
+**论文**：[arxiv:<id>] "<title>" (<venue/date>)
+**方法关键词**：<...>
+**GitLink**：search "<query>" → **0 命中**（查询串留底可复现）
+**GitHub 全球**：total_count = **2**（Top: <repo> <stars>⭐）→ 全球稀缺
+**机会建议**：<LLM 一句话：为何值得复现 + 开源到 GitLink>
+**起跑**：[按钮] fork 基准 <repo> → 创建 issue 粘论文 Algorithm 1 伪代码
+
+### 🧩 缺口 2 · 有需求无解答 [应用机会]
+**Issue**：<repo>#<n> "<subject>"（N 人讨论, open, <date>）
+**痛点**：<LLM 一句话研究性痛点归纳>
+**GitLink / GitHub**：均无成熟开源解
+**机会建议**：<LLM 一句话：可写应用级论文 + GitLink 落地>
+
+### ✅ 已诚实排除（非空白）
+- 论文 Y：GitLink 虽 0，但 GitHub 已 47 个实现（含官方）→ 全球已成熟，不报
+
+---
+<!-- gitlink-spark v1 | field:<field> | gaps:<N> | date:<YYYY-MM-DD> -->
+*由 gitlink-spark skill 生成。*
+````
+
+## 11. 起跑动作（M5）+ 安全护栏
+
+选定一张"理论无实现"缺口卡 → 用户确认 →
+1. `gitlink-cli repo +fork` 最近基准（GitHub Top 仓库 或 GitLink 最近实现）
+2. LLM 从 arXiv 论文抓取 Algorithm/Pseudocode 节
+3. `gitlink-cli issue +create` 在 fork 上建一个复现 todo issue，body 粘入论文伪代码 + 报告卡摘要
+
+**护栏**（沿用既有偏好）：默认预览确认；`--auto` 跳过预览但**永不 force-push、永不碰原仓库、永不自动 merge**；`--no-fork` 只出报告。
+
+## 12. 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| arXiv 返回空/超时 | 改 HTTPS 重试；仍空则报告"学术采失败"，降级用既有论文 |
+| `search +issues` 返回 HTML（已知坑） | 不用它；用 `issue +list` 逐仓库枚举 |
+| GitHub 未认证限流（10/min） | spark.py sleep ~7s 限速；建议设 `GITHUB_TOKEN` |
+| GitHub total_count 查询失败 | 该论文标注"GitHub 对照失败"，不进缺口判定 |
+| OpenAlex 503 | 跳过引用富集，不影响主报告 |
+| LLM 缺口无实证三件套 | 置信度门控丢弃 |
+| fork/issue 起跑失败 | 输出 fork 目标 + 伪代码文本供手动起跑 |
+| 二进制是 npm 旧版 | 强制 `./gitlink-cli` 或 `go build` |
+
+## 13. 验证计划（demo 2-3 例，GNN 领域）
+
+实跑 `spark.py --field "图神经网络"` + LLM 匹配，产出：
+- [ ] **≥1 张"理论无实现"缺口卡**（GitHub total_count < 10，全球稀缺），三件套可查
+- [ ] **≥1 张"需求无解答"缺口卡**（真实 open issue + 研究性痛点），三件套可查
+- [ ] **≥1 张"理论无实现"缺口卡走完 fork+issue 起跑**（截图/链接；demand 卡为写论文方向、无 fork 起跑）
+- [ ] "✅ 已诚实排除"区至少 1 条（GitHub ≥50 的非空白），证明阈值生效
+- [ ] 全报告无 LLM 编造（每条可溯源到 spark.py JSON）
+
+结果写入 `examples/spark-图神经网络.md`。
+
+## 14. 风险与未决
+
+- **arXiv 方法抽取**：从论文摘要自动抽"方法关键词"供 GitHub 查询，LLM 抽取有噪声 → spark.py 同时用论文标题关键词 + LLM 抽取双路查询 GitHub，取 total_count。
+- **GitLink GNN 仓库数量**：可能很少（之前扫到 GraphGallery 等）；需求侧缺口（issue）依赖仓库数，若太少则 demand 缺口样本不足 → 必要时放宽领域关键词（如含 `图`/`GNN`/`graph neural`）。
+- **GitHub 限流**：demo 一次 10 篇论文 × 1 查询 = 10 次，刚好未认证上限；建议跑 demo 时设 `GITHUB_TOKEN`。
+- **缺口"机会建议"主观**：靠实证三件套兜底；建议标注"机会仅为启发，需研究者自行判断"。
+
+## 15. 验收标准
+
+1. `skills/gitlink-spark/` 四件套齐全（SKILL.md + scripts/spark.py + REFERENCE.md + examples/）
+2. `spark.py` 可独立运行：`python spark.py --field 图神经网络` 输出合法融合 JSON
+3. GNN 跑出 2-3 张真实缺口卡（≥1 theory + ≥1 demand），每张三件套可查
+4. GitHub 阈值生效："已诚实排除"区至少 1 条
+5. ≥1 张缺口卡走完 fork+issue 起跑，有截图/链接
+6. 全报告每条缺口可溯源到 spark.py JSON（无 LLM 编造）
+7. REFERENCE.md 含缺口分类法 + 数据源实测结论 + 诚实边界
+8. README 登记 gitlink-spark

docs/superpowers/specs/2026-07-07-gitlink-research-fair-v2-design.md

@@ -0,0 +1,234 @@
+# gitlink-research-fair v2 设计（科研软件 X 光 / Research Software X-Ray）
+
+- **日期**：2026-07-07
+- **状态**：已批准，待编写实现计划
+- **作者**：CWQ + Claude
+- **定位**：对 v1（5 轴 FAIR 体检卡）的**重大升级**，回应"与 health 重合 / 报告千篇一律 / 假科研图谱 / 纯 Markdown 无脚本"四条批评。同一 skill 演进（保留 fair 名 + FAIR4RS 学术锚 + PR #5），重构内部。
+- **形态**：skill + **可运行 `scripts/fair.py`**（stdlib only），内容感知的科研分析工作流。
+
+---
+
+## 1. 背景与目标
+
+v1 的 fair 是"5 轴 FAIR 打分 + 报告卡"，存在四个真问题：
+1. 与 `gitlink-health` 形态重合（都是"聚合 gitlink-cli → AI → 打分报告"），且不够科研专属。
+2. 报告千篇一律（每仓库都是同一组 ✓/✗）；五轴 ASCII 雷达呈现烂；"科研关系图"是假的（只画作者/license/dataset?）。
+3. 纯 Markdown，无脚本，分析靠 LLM"看一眼"，不可复现。
+4. 缺真实科研洞察。
+
+v2 升级为**科研软件 X 光**：`fair.py` **真抽取** repo 内容（README/文件/依赖）→ 科研画像 JSON → LLM 对 4 个科研专属维度做裁决（每条带实证）→ **每 repo 特有的洞察报告**（裁决总览 + 真科研图谱 + 关键发现）+ 可选处方闭环。
+
+**目标**：内容感知（每份报告说本 repo 特有的东西）、科研专属（论文/数据/复现/引用，非通用软件质量）、可执行（fair.py 脚本）、demo 级（1-2 示例 + 1 答辩演示，非全场景覆盖）。
+
+## 2. 非目标（YAGNI）
+
+- 不做"实际运行代码验证复现"（静态复现就绪度检查即可，跑代码超 demo 范围）
+- 不做全场景/全领域覆盖——Feature_Critic + Edge 两个场景跑通即可
+- 不重复 spark（spark 跨仓跨源挖缺口；fair v2 单仓深挖科研产物）
+- 不自动 merge；处方默认预览
+- 不写 Go——skill + python 脚本
+- **砍掉 v1 的**：5 轴雷达、假 KG、OpenAlex 溯源（弱环节）、与 health 重合的通用质量味
+
+## 3. 关键决策（用户确认）
+
+| 决策点 | 选择 | 理由 |
+|--------|------|------|
+| 主角定位 | **科研软件 X 光**（深挖科研产物 + 4 维裁决） | 替代 5 轴打分，科研专属，每 repo 特有 |
+| 报告主视觉 | **混合**（裁决一行 + 真 Mermaid 图谱 + 关键发现） | 扫读 + 视觉兼顾，回应雷达烂 + 假图谱 + 千篇一律三连批 |
+| 改造方式 | **演进式**（保留 fair 名/FAIR4RS 锚/PR #5） | 不浪费 v1 资产，重构内部 |
+| 演示仓库 | **liyiying10/Feature_Critic**（ICML2019 paper-code） | README 明含 arxiv 链接，X 光论文溯源能真抽，4 维有区分度 |
+| 第二示例 | **Edgedev/Edge-Computing-Engine**（引擎类） | 与 Feature_Critic 不同 profile（license 冲突发现），证明不千篇一律 |
+| 脚本 | **fair.py**（stdlib only，fair.py 抽数据 + LLM 裁决） | 内容感知、可复现，取代 LLM"看一眼" |
+
+## 4. 与 v1 的差异 + 取舍
+
+| 维度 | v1 | v2 |
+|------|----|----|
+| 评分对象 | 5 轴 FAIR（F/A/I/R/Repro）通用软件质量 | 4 维科研专属（论文溯源/数据链/复现就绪/引用就绪） |
+| 数据来源 | LLM 看 gitlink-cli 元数据 | fair.py 真抽取 README/文件/依赖 |
+| 图谱 | 假 KG（作者/license/dataset?） | 真 Mermaid：paper↔method↔code↔dataset↔framework↔citation |
+| 报告 | 千篇一律 ✓/✗ + ASCII 雷达 | 裁决一行 + 真图谱 + **本 repo 特有关键发现** |
+| 脚本 | 无（纯 Markdown） | fair.py（可独立运行） |
+| 处方 | 有（保留） | 保留升级（CITATION.cff 从 README 抽 / requirements 从 import 扫） |
+| OpenAlex 溯源 | 有（弱、易 503） | **砍**（去掉更自洽） |
+| 落盘 | 有（v1 已补） | 保留（report-cards/<owner>-<repo>-xray.md） |
+
+## 5. 文件清单
+
+| 文件 | 动作 | 内容 |
+|------|------|------|
+| `skills/gitlink-research-fair/SKILL.md` | 重写 | X 光管道（抽数→裁决→图谱→报告→处方）、4 维裁决表、报告模板、真图谱说明、处方护栏、降级 |
+| `skills/gitlink-research-fair/scripts/fair.py` | 新增 | 真抽取（README/文件/依赖）→ 科研画像 JSON；stdlib only |
+| `skills/gitlink-research-fair/scripts/test_fair.py` | 新增 | 抽取函数单测（arxiv 正则 / 依赖检测 / 数据集识别 / 引用检测） |
+| `skills/gitlink-research-fair/REFERENCE.md` | 重写 | 4 维裁决细则 + 真 KG schema + 抽取规则 + 数据源实测 + 学术锚（FAIR4RS） |
+| `skills/gitlink-research-fair/examples/feature-critic-xray.md` | 新增 | Feature_Critic 真实 X 光走查（答辩演示脚本） |
+| `skills/gitlink-research-fair/examples/edge-xray.md` | 新增 | Edge 引擎类对照（license 冲突发现） |
+| `skills/README.md` / `skills/gitlink-workflow/SKILL.md` | 微调 | fair 描述改为"科研软件 X 光" |
+
+> v1 的旧 examples（songhui18 报告卡）替换为 v2 X 光示例。
+
+## 6. 命令接口
+
+### fair.py（可独立运行）
+```bash
+python skills/gitlink-research-fair/scripts/fair.py --owner liyiying10 --repo Feature_Critic
+# → stdout: 科研画像 JSON {paper, datasets, repro, citation, methods, frameworks, license, files, head_sha}
+```
+
+### skill 约定参数（SKILL.md）
+| 参数 | 默认 | 说明 |
+|------|------|------|
+| `--owner/--repo` | 自动从 cwd 解析 | 目标科研仓库（用 identifier） |
+| `--auto` | 关 | 跳过预览直接开处方 PR（仍受护栏） |
+| `--no-fork` | 关 | 只出 X 光报告，不开 PR |
+| `--refresh` | 关 | 即使有旧哨兵也重评 |
+
+## 7. 数据流（管道）
+
+```
+① 抽取（fair.py，确定性）
+   fetch readme (file +get) + file list (file +list) + repo +info + commit +list(HEAD sha)
+   → extract_paper / extract_datasets / assess_repro / assess_citation / extract_methods_frameworks
+   → 科研画像 JSON
+② 裁决（LLM，读 JSON）4 维 × (✅/⚠️/❌ + 具体证据)
+③ 真科研图谱（从画像生成 Mermaid）
+④ 渲染 X 光报告（hero：裁决一行 + 真图谱 + 关键发现 + 处方摘要 + 双裁决证书）→ 落盘
+⑤ 处方（可选）：fork → 建 CITATION.cff/requirements/Dockerfile → PR（默认预览）
+```
+
+**职责切分**：fair.py 只做**确定性抽取**（可复现）；**裁决与"关键发现"叙述**交给 LLM（语义），但每条必须引用 fair.py 抽到的实证。
+
+## 8. fair.py 抽取目标 + 科研画像 JSON schema
+
+抽取函数（纯函数单测 + gitlink-cli 网络层）：
+- `extract_paper(readme)` → 正则抽 arxiv id/URL、DOI、OpenReview/aclanthology、标题/作者/会议、"Code for the paper" 句式
+- `extract_datasets(readme, files)` → 已知数据集名（PACS/Cora/ImageNet/Visual Decathlon/MNIST/CIFAR…）+ 数据脚本（data_gen/get_data/download.sh）+ 数据 license/DOI
+- `assess_repro(files, readme)` → 依赖文件（requirements/go.mod/environment.yml/Dockerfile）+ 锁版本？；入口（main/train/run/*.py）；期望结果（results.md/accuracy/表）；环境说明
+- `assess_citation(files, readme)` → CITATION.cff / codemeta.json / .zenodo.json 存在？+ README BibTeX/引用文本
+- `extract_methods_frameworks(files, readme)` → 方法词 + 框架（torch/tensorflow/jax，从 import/文件名推断）
+
+科研画像 JSON：
+```json
+{
+  "repo": "liyiying10/Feature_Critic", "head_sha": "<sha>",
+  "paper": {"arxiv_id":"1901.11448","arxiv_url":"https://arxiv.org/abs/1901.11448","doi":null,
+            "title":"Feature-Critic Networks for Heterogeneous Domain Generalisation",
+            "authors":["Yiying Li","Yongxin Yang","Wei Zhou","Timothy M. Hospedales"],"venue":"ICML 2019","in_readme":true},
+  "datasets": [{"name":"PACS","evidence":"data_gen_PACS.py + README","download_script":"get_model_dataset.sh","license":null},
+               {"name":"Visual Decathlon","evidence":"data_gen_VD.py","download_script":null,"license":null}],
+  "repro": {"deps_files":[],"deps_pinned":false,"entry_points":["main_Feature_Critic.py","main_baseline.py"],
+            "expected_results":false,"env_spec":null},
+  "citation": {"cff":false,"codemeta":false,"zenodo":false,
+               "readme_bibtex":"Li, Yang, Zhou, Hospedales. Feature-Critic Networks... ICML 2019"},
+  "methods": ["feature-critic","meta-learning","domain generalisation"],
+  "frameworks": ["torch (inferred: alexnet/resnet/vggnet)"],
+  "license": {"file":false,"type":null,"conflict":null},
+  "files_count": 15
+}
+```
+
+## 9. 4 维裁决规则（LLM 读画像，每维 ✅/⚠️/❌ + 证据）
+
+| 维度 | ✅ | ⚠️ | ❌ |
+|------|---|----|----|
+| 论文溯源 | 有 arxiv/DOI + 元数据全 | 仅 README 文字提及，无稳定链接 | 无任何论文线索 |
+| 数据链 | 数据集命名 + 下载脚本 + license | 命名但无脚本/无 license | 未提及数据集 |
+| 复现就绪 | 依赖锁 + 入口 + 环境 + 期望结果四件齐全 | 有入口但缺依赖锁/环境/期望结果 | 无入口/无依赖 |
+| 引用就绪 | CITATION.cff/codemeta + DOI + 版本 | 仅 README 引用文本 | 无任何引用信息 |
+
+每维裁决**必须引用画像里的具体字段**（如"arxiv 1901.11448 已抽""requirements.txt 缺失"），无实证的判断丢弃。
+
+## 10. 真科研图谱 schema（Mermaid，从画像生成）
+
+节点（带状态色）：`Paper(arxiv+venue)` / `Method` / `Code(file)` / `Dataset` / `Framework` / `Citation`
+边：`Paper —proposes→ Method`、`Method —implements→ Code`、`Code —uses→ Dataset`、`Code —depends→ Framework`、`Paper —cited-via→ Citation`
+状态色：✅绿 / ⚠️黄 / ❌红（缺失项）。
+
+Feature_Critic 实例化：`ICML2019(1901.11448)✅ —proposes→ Feature-Critic方法 —implements→ main_Feature_Critic.py✅ —uses→ PACS⚠️/Visual Decathlon⚠️ —depends→ PyTorch(推断)⚠️；Paper —cited-via→ 无CITATION❌`。真有科研含义，非 v1 假图。
+
+## 11. 报告格式（混合主视觉，hero）
+
+````markdown
+🔬 **科研软件 X 光 — liyiying10/Feature_Critic**
+
+═══════════════════════════════════════
+论文溯源 ✅ ｜ 数据链 ⚠️ ｜ 复现就绪 ⚠️ ｜ 引用就绪 ❌
+═══════════════════════════════════════
+
+### 🧬 真科研图谱
+```mermaid
+graph LR
+  P[ICML2019 arxiv:1901.11448]:::ok -->|proposes| M[Feature-Critic 方法]
+  M -->|implements| C[main_Feature_Critic.py]:::ok
+  C -->|uses| D1[PACS]:::warn
+  C -->|uses| D2[Visual Decathlon]:::warn
+  C -->|depends| F[PyTorch 推断]:::warn
+  P -->|cited-via| Ci[无 CITATION.cff]:::bad
+  classDef ok fill:#cfe,stroke:#3a3; classDef warn fill:#ffe,stroke:#cc3; classDef bad fill:#fee,stroke:#c33;
+```
+
+### 🔍 关键发现（本 repo 特有）
+- 论文 ICML2019 arxiv:1901.11448 已溯源 ✓，但**无 CITATION.cff** → 机器不可引用
+- 数据集 PACS/Visual Decathlon 命名 + `get_model_dataset.sh`，但**无数据 license**
+- 入口 `main_Feature_Critic.py` 在，但**无 requirements.txt** → 依赖未锁，复现风险
+- 框架：alexnet/resnet/vggnet（疑似 PyTorch，**依赖未声明**）
+
+### 🔧 处方（可选）
+补 `CITATION.cff`（从 README 抽的引用）+ `requirements.txt`（从 import 扫）+ `Dockerfile`
+
+### 📜 双裁决证书
+复现就绪 ⚠️ 部分 ｜ 引用就绪 ❌ ｜ 锚定 commit `<sha>`
+
+---
+<!-- gitlink-research-fair v2 | repo:<owner>/<repo> | paper:? | repro:? | cite:? | sha:<head> -->
+*由 gitlink-research-fair v2（科研软件 X 光）生成。*
+````
+
+报告**始终落盘** `report-cards/<owner>-<repo>-xray.md`（沿用 v1 落盘原则）。
+
+## 12. 处方闭环 + 护栏（保留 v1 升级）
+
+对 ❌/⚠️ 项生成修复：`CITATION.cff`（从 README 抽的引用文本构造）+ `requirements.txt`（从代码 import 扫出依赖）+ `Dockerfile`（模板）。→ fork → PR（默认预览，`--auto` 跳过；永不 force-push/碰原仓库/自动 merge；`--no-fork` 报告已落盘）。
+
+## 13. 错误处理与降级
+
+| 情况 | 处理 |
+|------|------|
+| README 读取失败 | 降级用 file list + 元数据，标注"README 不可读，结论受限" |
+| `file +list` data 为字符串 | json.loads 解套（已知结构） |
+| arxiv/DOI 抽不到 | 论文溯源维度判 ⚠️ 或 ❌，据实 |
+| 框架无法推断 | frameworks 标 "未知"，不编造 |
+| `--repo` 用了中文显示名 404 | 提示用 identifier（从 search 取） |
+| fork/PR 失败 | 处方物料落本地，告知路径 |
+| 报告/PR 发布失败 | 报告已在 ④ 落盘，告知路径 |
+
+## 14. 验证计划（1-2 示例 + 答辩演示）
+
+- **答辩现场演示 + 示例1**：`liyiying10/Feature_Critic`
+  - [ ] fair.py 抽出 arxiv 1901.11448 + ICML2019 + PACS/VD 数据集 + 入口 + 无 CITATION/requirements
+  - [ ] 4 维裁决：论文 ✅ / 数据 ⚠️ / 复现 ⚠️ / 引用 ❌
+  - [ ] 真图谱画出 6 类节点 + 状态色
+  - [ ] 关键发现 ≥3 条本 repo 特有
+- **示例2**：`Edgedev/Edge-Computing-Engine`
+  - [ ] X 光发现 license 冲突（Apache vs README"禁商用"）作为关键发现
+  - [ ] 与 Feature_Critic 报告内容明显不同（证明不千篇一律）
+- 两示例均落盘 report-cards/ + 写入 examples/。
+
+## 15. 风险与未决
+
+- **README 抽取覆盖度**：正则抽 arxiv/DOI 依赖 README 写法；对非标准 README 可能漏。fair.py 同时匹配多种句式（arxiv URL / arXiv:id / 裸 id）兜底。
+- **框架推断**：从文件名/import 推断 PyTorch/TF 有噪声 → 标"推断"，不肯定。
+- **数据集识别**：已知数据集名白名单有限；未知名标"unnamed dataset (mentioned)"。
+- **GitLink file +list data 字符串化**：fair.py 需 json.loads 解套（已知）。
+- **演示仓库稳定性**：Feature_Critic 是 1⭐ 小仓但 ICML2019 真实学术代码，内容稳定；演示前复跑一次确认 README/文件未变。
+
+## 16. 验收标准
+
+1. `skills/gitlink-research-fair/` 含 SKILL.md（重写）+ scripts/fair.py + test_fair.py + REFERENCE.md（重写）+ 2 个 examples
+2. `python fair.py --owner liyiying10 --repo Feature_Critic` 输出合法科研画像 JSON（含真实 arxiv 1901.11448）
+3. test_fair.py 抽取函数单测全过
+4. Feature_Critic X 光：4 维裁决正确（论文✅/数据⚠️/复现⚠️/引用❌）+ 真图谱 ≥6 节点 + 关键发现 ≥3 条
+5. Edge X 光：license 冲突作为关键发现，报告与 Feature_Critic 明显不同
+6. 报告落盘 report-cards/
+7. v1 的 5 轴雷达/假 KG 已从 SKILL.md 移除
+8. README/workflow 描述更新为"科研软件 X 光"

internal/config/config.go

@@ -14,10 +14,11 @@ const (
 )
 
 type Config struct {
-	BaseURL string `yaml:"base_url"`
-	Format  string `yaml:"default_format"`
-	Editor  string `yaml:"editor,omitempty"`
-	Pager   string `yaml:"pager,omitempty"`
+	BaseURL        string `yaml:"base_url"`
+	Format         string `yaml:"default_format"`
+	Editor         string `yaml:"editor,omitempty"`
+	Pager          string `yaml:"pager,omitempty"`
+	DeepSeekAPIKey string `yaml:"deepseek_api_key,omitempty"`
 }
 
 func DefaultConfig() *Config {
@@ -91,6 +92,8 @@ func Get(key string) (string, error) {
 		return cfg.Editor, nil
 	case "pager":
 		return cfg.Pager, nil
+	case "deepseek_api_key":
+		return cfg.DeepSeekAPIKey, nil
 	default:
 		return "", nil
 	}
@@ -110,6 +113,8 @@ func Set(key, value string) error {
 		cfg.Editor = value
 	case "pager":
 		cfg.Pager = value
+	case "deepseek_api_key":
+		cfg.DeepSeekAPIKey = value
 	}
 	return Save(cfg)
 }

internal/output/formatter.go

@@ -66,7 +66,19 @@ func printTable(w io.Writer, envelope *Envelope) error {
 		return nil
 	}
 
-	switch data := envelope.Data.(type) {
+	// Convert struct types to generic map/slice via JSON round-trip
+	data := envelope.Data
+	if _, ok := data.(map[string]interface{}); !ok {
+		if _, ok := data.([]interface{}); !ok {
+			generic, err := toGeneric(data)
+			if err != nil {
+				return printJSON(w, envelope)
+			}
+			data = generic
+		}
+	}
+
+	switch data := data.(type) {
 	case []interface{}:
 		return printSliceTable(w, data)
 	case map[string]interface{}:
@@ -82,11 +94,23 @@ func printTable(w io.Writer, envelope *Envelope) error {
 	}
 }
 
+func toGeneric(v interface{}) (interface{}, error) {
+	b, err := json.Marshal(v)
+	if err != nil {
+		return nil, err
+	}
+	var result interface{}
+	if err := json.Unmarshal(b, &result); err != nil {
+		return nil, err
+	}
+	return result, nil
+}
+
 func findSliceInMap(m map[string]interface{}) []interface{} {
 	for _, key := range []string{
 		"issues", "pull_requests", "milestones", "webhooks", "issue_tags",
 		"commits", "files", "members", "collaborators", "users", "branches",
-		"releases", "entries", "tags", "watchers",
+		"releases", "entries", "tags", "watchers", "results",
 	} {
 		if v, ok := m[key]; ok {
 			if slice, ok := v.([]interface{}); ok && len(slice) > 0 {

main.go

@@ -11,3 +11,4 @@ func main() {
 		os.Exit(1)
 	}
 }
+// test: code-quality watch trigger test 1782976608

pr-test-file.txt

@@ -1 +0,0 @@
-PR Test 2026年 4月 7日 星期二 11时45分56秒 CST

scripts/setup-skills.sh

@@ -0,0 +1,79 @@
+#!/bin/bash
+# Link skills/gitlink-* into ~/.claude/skills/ so Claude Code can load and
+# invoke them through the Skill tool (e.g. `Skill gitlink-issue`).
+#
+# Cross-platform:
+#   - Windows (Git Bash / MSYS): directory junction (mklink /J, no admin needed)
+#   - macOS / Linux: symlink
+#
+# Links point at the in-repo skills/ directory, so updating the repo keeps the
+# Skill content in sync. The script is idempotent and safe to re-run.
+#
+# CRITICAL (Windows): an existing link is removed with `rmdir` (no /s), never
+# `rm -rf` — `rm -rf` would follow the junction and DELETE THE SOURCE FILES.
+
+set -e
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+PROJECT_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
+SRC="$PROJECT_DIR/skills"
+DST="$HOME/.claude/skills"
+
+# --- collect gitlink-* skills (top-level skills/ + workflow-bundled skills) ---
+shopt -s nullglob
+SKILLS=("$SRC"/gitlink-*/ "$PROJECT_DIR/shortcuts/workflow/skills"/gitlink-*/)
+shopt -u nullglob
+if [ ${#SKILLS[@]} -eq 0 ]; then
+  echo "No gitlink-* skills found under $SRC" >&2
+  exit 1
+fi
+
+mkdir -p "$DST"
+
+# --- detect platform ---
+case "$(uname -s)" in
+  MINGW*|MSYS*|CYGWIN*) PLATFORM=windows ;;
+  *)                    PLATFORM=unix ;;
+esac
+
+OK=0
+FAIL=0
+
+for skill in "${SKILLS[@]}"; do
+  skill="${skill%/}"                      # strip trailing slash for clean paths
+  name="$(basename "$skill")"
+  link="$DST/$name"
+
+  if [ "$PLATFORM" = "windows" ]; then
+    win_link="$(cygpath -w "$link")"
+    win_src="$(cygpath -w "$skill")"
+
+    if [ -e "$link" ] || [ -L "$link" ]; then
+      # rmdir (no /s) removes only the junction/symlink, never the target.
+      if ! cmd //c rmdir "$win_link" >/dev/null 2>&1; then
+        echo "SKIP  $name  (existing path is not a link — left untouched)"
+        FAIL=$((FAIL+1)); continue
+      fi
+    fi
+
+    if powershell -NoProfile -Command \
+      "New-Item -ItemType Junction -Path '$win_link' -Target '$win_src' -ErrorAction Stop" \
+      >/dev/null 2>&1; then
+      echo "OK    $name"; OK=$((OK+1))
+    else
+      echo "FAIL  $name"; FAIL=$((FAIL+1))
+    fi
+  else
+    # ln -sfn replaces an existing symlink safely (does not follow it).
+    if ln -sfn "$skill" "$link"; then
+      echo "OK    $name"; OK=$((OK+1))
+    else
+      echo "FAIL  $name"; FAIL=$((FAIL+1))
+    fi
+  fi
+done
+
+echo ""
+echo "Done: $OK linked, $FAIL failed."
+echo "Target: $DST"
+echo "Skills are now invocable via the Claude Code Skill tool."

shortcuts/common/types.go

@@ -36,6 +36,7 @@ type RuntimeContext struct {
 	Repo   string
 	Format string
 	Args   map[string]string
+	AIMode string
 }
 
 // NewRuntimeContext creates a RuntimeContext with auto-resolved owner/repo.

shortcuts/issue/batch.go

@@ -280,9 +280,9 @@ func assignIssue(ctx *common.RuntimeContext, number, user string) error {
 	}
 
 	body := map[string]interface{}{
-		"subject":        current.Subject,
-		"description":    current.Description,
-		"assigned_to_id": userID,
+		"subject":      current.Subject,
+		"description":  current.Description,
+		"assigner_ids": []int64{int64(userID)},
 	}
 	if current.StatusID != nil {
 		body["status_id"] = current.StatusID
@@ -474,6 +474,89 @@ func setMilestone(ctx *common.RuntimeContext, number, milestoneID string) error
 	return nil
 }
 
+func newBatchCommentShortcut() *common.Shortcut {
+	return &common.Shortcut{
+		Name:        "batch-comment",
+		Description: "Add a comment to multiple issues by issue numbers or a CSV file",
+		Flags: []common.Flag{
+			{Name: "numbers", Short: "n", Usage: "Comma-separated issue numbers from the web URL, for example: 1,2,3"},
+			{Name: "from", Usage: "Read issue numbers from a CSV file. Supports a number/issue_number/project_issues_index column or first column without header"},
+			{Name: "body", Short: "b", Usage: "Comment body", Required: true},
+			{Name: "dry-run", Usage: "Preview the issues that would be commented on without changing them", Bool: true, Default: "false"},
+		},
+		Run: runBatchComment,
+	}
+}
+
+func runBatchComment(ctx *common.RuntimeContext) error {
+	if err := ctx.ResolveOwnerRepo(); err != nil {
+		return fmt.Errorf("解析仓库信息失败: %w", err)
+	}
+
+	body, err := ctx.RequireArg("body")
+	if err != nil {
+		return err
+	}
+
+	start := time.Now()
+
+	numbers, err := collectIssueNumbers(ctx.Arg("numbers"), ctx.Arg("from"))
+	if err != nil {
+		return err
+	}
+	if len(numbers) == 0 {
+		return fmt.Errorf("未提供 Issue 编号，请使用 --numbers 1,2,3 或 --from issues.csv")
+	}
+
+	dryRun := parseBool(ctx.Arg("dry-run"))
+	summary := batchSummary{
+		Repository: fmt.Sprintf("%s/%s", ctx.Owner, ctx.Repo),
+		DryRun:     dryRun,
+		Total:      len(numbers),
+		Results:    make([]batchResult, 0, len(numbers)),
+	}
+
+	for _, number := range numbers {
+		result := batchResult{Number: number, Action: "comment"}
+		if dryRun {
+			result.Status = "planned"
+			summary.Succeeded++
+			summary.Results = append(summary.Results, result)
+			continue
+		}
+
+		if err := commentIssue(ctx, number, body); err != nil {
+			result.Status = "failed"
+			result.Error = err.Error()
+			summary.Failed++
+		} else {
+			result.Status = "commented"
+			summary.Succeeded++
+		}
+		summary.Results = append(summary.Results, result)
+	}
+
+	summary.Duration = time.Since(start).String()
+
+	if err := ctx.OutputData(summary); err != nil {
+		return err
+	}
+	if summary.Failed > 0 {
+		return fmt.Errorf("%d / %d 个 Issue 评论失败", summary.Failed, summary.Total)
+	}
+	return nil
+}
+
+func commentIssue(ctx *common.RuntimeContext, number, body string) error {
+	payload := map[string]interface{}{
+		"notes": body,
+	}
+	if _, err := ctx.CallAPI("POST", fmt.Sprintf("%s/issues/%s/journals", v1RepoPath(ctx), number), payload); err != nil {
+		return fmt.Errorf("添加评论: %w", err)
+	}
+	return nil
+}
+
 func parseCommaSeparated(value string) []string {
 	if strings.TrimSpace(value) == "" {
 		return nil

shortcuts/issue/issue.go

@@ -28,6 +28,7 @@ func Shortcuts() []*common.Shortcut {
 		newBatchAssignShortcut(),
 		newBatchLabelShortcut(),
 		newBatchMilestoneShortcut(),
+		newBatchCommentShortcut(),
 		{
 			Name:        "list",
 			Description: "List issues",
@@ -257,9 +258,9 @@ func Shortcuts() []*common.Shortcut {
 					return fmt.Errorf("user 参数必须是数字 ID，而不是用户名")
 				}
 				body := map[string]interface{}{
-					"subject":        current.Subject,
-					"description":    current.Description,
-					"assigned_to_id": userID,
+					"subject":     current.Subject,
+					"description": current.Description,
+					"assigner_ids": []int64{int64(userID)},
 				}
 				if current.StatusID != nil {
 					body["status_id"] = current.StatusID

shortcuts/issue/issue_test.go

@@ -281,16 +281,17 @@ func TestIssueComment(t *testing.T) {
 }
 
 func TestIssueAssignSendsCorrectUser(t *testing.T) {
-	var assignPath string
 	var assignPayload map[string]interface{}
 	server := newIssueTestServer(t, func(w http.ResponseWriter, r *http.Request) {
 		switch {
-		case r.Method == "PUT" && r.URL.Path == "/v1/owner/repo/issues/42/assignees.json":
-			assignPath = r.URL.Path
-			assignPayload = decodeJSON(t, r)
+		case r.Method == "GET" && r.URL.Path == "/v1/owner/repo/issues/42.json":
 			writeJSON(t, w, map[string]interface{}{
-				"message": "指派成功",
+				"subject":     "Existing title",
+				"description": "Existing description",
 			})
+		case r.Method == "PATCH" && r.URL.Path == "/v1/owner/repo/issues/42.json":
+			assignPayload = decodeJSON(t, r)
+			writeJSON(t, w, map[string]interface{}{"message": "指派成功"})
 		default:
 			t.Fatalf("unexpected request: %s %s", r.Method, r.URL.Path)
 		}
@@ -299,16 +300,18 @@ func TestIssueAssignSendsCorrectUser(t *testing.T) {
 
 	err := runIssueShortcut(t, server, "assign", map[string]string{
 		"number": "42",
-		"user":   "zhangsan",
+		"user":   "42",
 	})
 	if err != nil {
 		t.Fatalf("assign shortcut failed: %v", err)
 	}
 
-	if assignPath == "" {
-		t.Fatal("assign endpoint was not called")
+	// GitLink 责任人字段是 assigner_ids（数组），不是 assigned_to_id
+	got, ok := assignPayload["assigner_ids"].([]interface{})
+	if !ok || len(got) != 1 {
+		t.Fatalf("expected assigner_ids [42], got %#v", assignPayload["assigner_ids"])
 	}
-	assertEqual(t, assignPayload["assigned_to_id"], "zhangsan")
+	assertEqual(t, got[0], float64(42))
 }
 
 func TestIssueAssignRequiresNumberAndUser(t *testing.T) {
@@ -386,15 +389,15 @@ func TestIssueLabelRemoveSendsDeleteRequests(t *testing.T) {
 }
 
 func TestBatchAssign(t *testing.T) {
-	var assignPath string
 	var assignPayload map[string]interface{}
 	server := newIssueTestServer(t, func(w http.ResponseWriter, r *http.Request) {
 		switch {
-		case r.Method == "PUT" && r.URL.Path == "/v1/owner/repo/issues/1/assignees.json":
-			assignPath = r.URL.Path
+		case r.Method == "GET" && (r.URL.Path == "/v1/owner/repo/issues/1.json" || r.URL.Path == "/v1/owner/repo/issues/2.json"):
+			writeJSON(t, w, map[string]interface{}{"subject": "title", "description": "desc"})
+		case r.Method == "PATCH" && r.URL.Path == "/v1/owner/repo/issues/1.json":
 			assignPayload = decodeJSON(t, r)
 			writeJSON(t, w, map[string]interface{}{"message": "指派成功"})
-		case r.Method == "PUT" && r.URL.Path == "/v1/owner/repo/issues/2/assignees.json":
+		case r.Method == "PATCH" && r.URL.Path == "/v1/owner/repo/issues/2.json":
 			writeJSON(t, w, map[string]interface{}{"message": "指派成功"})
 		default:
 			t.Fatalf("unexpected request: %s %s", r.Method, r.URL.Path)
@@ -404,17 +407,18 @@ func TestBatchAssign(t *testing.T) {
 
 	err := runIssueShortcut(t, server, "batch-assign", map[string]string{
 		"numbers": "1,2",
-		"user":    "zhangsan",
+		"user":    "42",
 		"dry-run": "false",
 	})
 	if err != nil {
 		t.Fatalf("batch-assign failed: %v", err)
 	}
 
-	if assignPath == "" {
-		t.Fatal("assign endpoint was not called")
+	got, ok := assignPayload["assigner_ids"].([]interface{})
+	if !ok || len(got) != 1 {
+		t.Fatalf("expected assigner_ids [42], got %#v", assignPayload["assigner_ids"])
 	}
-	assertEqual(t, assignPayload["assigned_to_id"], "zhangsan")
+	assertEqual(t, got[0], float64(42))
 }
 
 func TestBatchAssignDryRun(t *testing.T) {

shortcuts/member/batch.go

@@ -0,0 +1,204 @@
+package member
+
+import (
+	"encoding/csv"
+	"fmt"
+	"os"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+)
+
+type batchAddResult struct {
+	User   string `json:"user" yaml:"user"`
+	Action string `json:"action" yaml:"action"`
+	Status string `json:"status" yaml:"status"`
+	Error  string `json:"error,omitempty" yaml:"error,omitempty"`
+}
+
+type batchAddSummary struct {
+	Owner     string            `json:"owner" yaml:"owner"`
+	Repo      string            `json:"repo" yaml:"repo"`
+	DryRun    bool              `json:"dry_run" yaml:"dry_run"`
+	Total     int               `json:"total" yaml:"total"`
+	Succeeded int               `json:"succeeded" yaml:"succeeded"`
+	Failed    int               `json:"failed" yaml:"failed"`
+	Duration  string            `json:"duration" yaml:"duration"`
+	Results   []batchAddResult  `json:"results" yaml:"results"`
+}
+
+func batchAddShortcut() *common.Shortcut {
+	return &common.Shortcut{
+		Name:        "batch-add",
+		Description: "批量添加成员到项目，支持逗号分隔列表或 CSV 文件",
+		Flags: []common.Flag{
+			{Name: "users", Short: "u", Usage: "逗号分隔的用户数字 ID，例如: 42,99,105"},
+			{Name: "from", Usage: "从 CSV 文件读取用户 ID。支持 user_id/id/user 列名或无表头首列"},
+			{Name: "dry-run", Usage: "仅预览将要添加的成员，不实际执行", Bool: true, Default: "false"},
+		},
+		Run: runBatchAdd,
+	}
+}
+
+func runBatchAdd(ctx *common.RuntimeContext) error {
+	start := time.Now()
+
+	if err := ctx.ResolveOwnerRepo(); err != nil {
+		return err
+	}
+
+	userIDs, err := collectUserIDs(ctx.Arg("users"), ctx.Arg("from"))
+	if err != nil {
+		return err
+	}
+	if len(userIDs) == 0 {
+		return fmt.Errorf("未提供用户 ID，请使用 --users 42,99 或 --from users.csv")
+	}
+
+	dryRun := parseMemberBool(ctx.Arg("dry-run"))
+
+	summary := batchAddSummary{
+		Owner:   ctx.Owner,
+		Repo:    ctx.Repo,
+		DryRun:  dryRun,
+		Total:   len(userIDs),
+		Results: make([]batchAddResult, 0, len(userIDs)),
+	}
+
+	for _, uid := range userIDs {
+		result := batchAddResult{User: uid, Action: "add"}
+		if dryRun {
+			result.Status = "planned"
+			summary.Succeeded++
+			summary.Results = append(summary.Results, result)
+			continue
+		}
+
+		id, _ := strconv.ParseInt(uid, 10, 64)
+		body := map[string]interface{}{"user_id": id}
+		if _, err := ctx.CallAPI("POST", ctx.RepoPath()+"/collaborators", body); err != nil {
+			result.Status = "failed"
+			result.Error = err.Error()
+			summary.Failed++
+		} else {
+			result.Status = "added"
+			summary.Succeeded++
+		}
+		summary.Results = append(summary.Results, result)
+	}
+
+	summary.Duration = time.Since(start).String()
+
+	if err := ctx.OutputData(summary); err != nil {
+		return err
+	}
+	if summary.Failed > 0 {
+		return fmt.Errorf("%d / %d 个成员添加失败", summary.Failed, summary.Total)
+	}
+	return nil
+}
+
+func collectUserIDs(usersValue, csvPath string) ([]string, error) {
+	ids, err := parseUserIDList(usersValue)
+	if err != nil {
+		return nil, err
+	}
+	if csvPath == "" {
+		return ids, nil
+	}
+
+	csvIDs, err := readUserIDsFromCSV(csvPath)
+	if err != nil {
+		return nil, err
+	}
+	return mergeUserIDLists(ids, csvIDs), nil
+}
+
+func parseUserIDList(value string) ([]string, error) {
+	if strings.TrimSpace(value) == "" {
+		return nil, nil
+	}
+	return normalizeUserIDs(strings.Split(value, ","))
+}
+
+func readUserIDsFromCSV(path string) ([]string, error) {
+	file, err := os.Open(path)
+	if err != nil {
+		return nil, fmt.Errorf("读取 CSV 文件失败: %w", err)
+	}
+	defer file.Close()
+
+	reader := csv.NewReader(file)
+	reader.TrimLeadingSpace = true
+	records, err := reader.ReadAll()
+	if err != nil {
+		return nil, fmt.Errorf("解析 CSV 文件失败: %w", err)
+	}
+	if len(records) == 0 {
+		return nil, nil
+	}
+
+	idCol := -1
+	startRow := 0
+	for i, cell := range records[0] {
+		switch strings.ToLower(strings.TrimSpace(cell)) {
+		case "user_id", "id", "user", "uid":
+			idCol = i
+			startRow = 1
+		}
+	}
+	if idCol == -1 {
+		idCol = 0
+	}
+
+	values := make([]string, 0, len(records)-startRow)
+	for _, record := range records[startRow:] {
+		if idCol >= len(record) {
+			continue
+		}
+		values = append(values, record[idCol])
+	}
+	return normalizeUserIDs(values)
+}
+
+func normalizeUserIDs(values []string) ([]string, error) {
+	ids := make([]string, 0, len(values))
+	seen := map[string]bool{}
+	for _, value := range values {
+		id := strings.TrimSpace(value)
+		if id == "" {
+			continue
+		}
+		if _, err := strconv.ParseInt(id, 10, 64); err != nil {
+			return nil, fmt.Errorf("无效的用户 ID %q: 必须是整数", id)
+		}
+		if seen[id] {
+			continue
+		}
+		seen[id] = true
+		ids = append(ids, id)
+	}
+	return ids, nil
+}
+
+func mergeUserIDLists(values ...[]string) []string {
+	merged := []string{}
+	seen := map[string]bool{}
+	for _, ids := range values {
+		for _, id := range ids {
+			if seen[id] {
+				continue
+			}
+			seen[id] = true
+			merged = append(merged, id)
+		}
+	}
+	return merged
+}
+
+func parseMemberBool(value string) bool {
+	parsed, err := strconv.ParseBool(strings.TrimSpace(value))
+	return err == nil && parsed
+}

shortcuts/member/member.go

@@ -10,6 +10,7 @@ import (
 
 func Shortcuts() []*common.Shortcut {
 	return []*common.Shortcut{
+		batchAddShortcut(),
 		{
 			Name:        "list",
 			Description: "List project members",

shortcuts/register.go

@@ -22,6 +22,9 @@ import (
 	"github.com/gitlink-org/gitlink-cli/shortcuts/watch"
 	"github.com/gitlink-org/gitlink-cli/shortcuts/webhook"
 	"github.com/gitlink-org/gitlink-cli/shortcuts/wiki"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/cli"
+	_ "github.com/gitlink-org/gitlink-cli/shortcuts/workflow/defs"
+	_ "github.com/gitlink-org/gitlink-cli/shortcuts/workflow/rules"
 )
 
 // RegisterAll mounts all shortcut groups onto the root command.
@@ -45,6 +48,7 @@ func RegisterAll(root *cobra.Command) {
 		"watch":     watch.Shortcuts(),
 		"star":      star.Shortcuts(),
 		"wiki":      wiki.Shortcuts(),
+		"workflow":  cli.Shortcuts(),
 	}
 
 	descriptions := map[string]string{
@@ -66,6 +70,7 @@ func RegisterAll(root *cobra.Command) {
 		"watch":     "Watch repository operations",
 		"star":      "Star repository operations",
 		"wiki":      "Wiki page operations",
+		"workflow":  "Automated workflow operations",
 	}
 
 	for name, shortcuts := range groups {

shortcuts/repo/batch.go

@@ -25,20 +25,22 @@ type batchRepoSummary struct {
 	Results   []batchRepoResult `json:"results" yaml:"results"`
 }
 
-func newBatchDeleteShortcut() *common.Shortcut {
+func newBatchCreateShortcut() *common.Shortcut {
 	return &common.Shortcut{
-		Name:        "batch-delete",
-		Description: "批量删除仓库，支持逗号分隔列表或 CSV 文件",
+		Name:        "batch-create",
+		Description: "批量创建仓库，支持逗号分隔列表或 CSV 文件",
 		Flags: []common.Flag{
-			{Name: "repos", Short: "r", Usage: "逗号分隔的仓库标识，格式为 owner/repo，例如: alice/proj1,bob/proj2"},
-			{Name: "from", Usage: "从 CSV 文件读取仓库标识。支持 owner/repo、owner,repo 双列或 owner/repo 单列格式"},
-			{Name: "dry-run", Usage: "仅预览将要删除的仓库，不实际执行", Bool: true, Default: "false"},
+			{Name: "repos", Short: "r", Usage: "逗号分隔的仓库名称，例如: repo1,repo2,repo3"},
+			{Name: "from", Usage: "从 CSV 文件读取仓库名称。支持 name/repo/repository 列名或无表头首列"},
+			{Name: "description", Short: "d", Usage: "仓库描述（所有仓库共用一个描述）"},
+			{Name: "private", Usage: "设为私有仓库 (true/false)", Default: "false"},
+			{Name: "dry-run", Usage: "仅预览将要创建的仓库，不实际执行", Bool: true, Default: "false"},
 		},
-		Run: runBatchDelete,
+		Run: runBatchCreate,
 	}
 }
 
-func runBatchDelete(ctx *common.RuntimeContext) error {
+func runBatchCreate(ctx *common.RuntimeContext) error {
 	start := time.Now()
 
 	repos, err := collectRepos(ctx.Arg("repos"), ctx.Arg("from"))
@@ -46,24 +48,44 @@ func runBatchDelete(ctx *common.RuntimeContext) error {
 		return err
 	}
 	if len(repos) == 0 {
-		return fmt.Errorf("未提供仓库标识，请使用 --repos owner/repo1,owner/repo2 或 --from repos.csv")
+		return fmt.Errorf("未提供仓库名称，请使用 --repos repo1,repo2 或 --from repos.csv")
+	}
+
+	// 仅取仓库名，不需要 owner/repo 格式
+	names := make([]string, len(repos))
+	for i, r := range repos {
+		parts := strings.SplitN(r, "/", 2)
+		names[i] = parts[len(parts)-1]
 	}
 
 	dryRun := parseBool(ctx.Arg("dry-run"))
-	dryRunVal := false
-	if ctx.Arg("dry-run") != "" {
-		dryRunVal = dryRun
-	}
-	_ = dryRunVal
-
 	summary := batchRepoSummary{
-		Total:   len(repos),
-		Results: make([]batchRepoResult, 0, len(repos)),
+		Total:   len(names),
+		Results: make([]batchRepoResult, 0, len(names)),
 	}
 
-	for _, repoID := range repos {
-		parts := strings.SplitN(repoID, "/", 2)
-		result := batchRepoResult{Repo: repoID, Action: "delete"}
+	var userLogin string
+	var userID int
+	if !dryRun {
+		userEnv, err := ctx.CallAPI("GET", "/users/me", nil)
+		if err != nil {
+			return fmt.Errorf("获取当前用户信息失败: %w", err)
+		}
+		userData, _ := userEnv.Data.(map[string]interface{})
+		login, _ := userData["login"].(string)
+		if login == "" {
+			return fmt.Errorf("无法获取当前用户名")
+		}
+		userLogin = login
+		uid, _ := userData["user_id"].(float64)
+		userID = int(uid)
+	}
+
+	private := ctx.Arg("private") == "true"
+	desc := ctx.Arg("description")
+
+	for _, name := range names {
+		result := batchRepoResult{Repo: name, Action: "create"}
 		if dryRun {
 			result.Status = "planned"
 			summary.Succeeded++
@@ -71,13 +93,25 @@ func runBatchDelete(ctx *common.RuntimeContext) error {
 			continue
 		}
 
-		path := fmt.Sprintf("/%s/%s", parts[0], parts[1])
-		if _, err := ctx.CallAPI("DELETE", path, nil); err != nil {
+		body := map[string]interface{}{
+			"name":            name,
+			"repository_name": name,
+			"user_id":         userID,
+		}
+		if desc != "" {
+			body["description"] = desc
+		}
+		if private {
+			body["private"] = true
+		}
+
+		path := fmt.Sprintf("/%s/%s", userLogin, name)
+		if _, err := ctx.CallAPI("POST", path, body); err != nil {
 			result.Status = "failed"
 			result.Error = err.Error()
 			summary.Failed++
 		} else {
-			result.Status = "deleted"
+			result.Status = "created"
 			summary.Succeeded++
 		}
 		summary.Results = append(summary.Results, result)
@@ -89,7 +123,7 @@ func runBatchDelete(ctx *common.RuntimeContext) error {
 		return err
 	}
 	if summary.Failed > 0 {
-		return fmt.Errorf("%d / %d 个仓库删除失败", summary.Failed, summary.Total)
+		return fmt.Errorf("%d / %d 个仓库创建失败", summary.Failed, summary.Total)
 	}
 	return nil
 }
@@ -100,7 +134,7 @@ func newBatchForkShortcut() *common.Shortcut {
 		Description: "批量 Fork 仓库，支持逗号分隔列表或 CSV 文件",
 		Flags: []common.Flag{
 			{Name: "repos", Short: "r", Usage: "逗号分隔的仓库标识，格式为 owner/repo，例如: alice/proj1,bob/proj2"},
-			{Name: "from", Usage: "从 CSV 文件读取仓库标识。支持 owner/repo、owner,repo 双列或 owner/repo 单列格式"},
+			{Name: "from", Usage: "从 CSV 文件读取仓库标识。支持 owner/repo 单列或 owner、repo 双列格式"},
 			{Name: "dry-run", Usage: "仅预览将要 Fork 的仓库，不实际执行", Bool: true, Default: "false"},
 		},
 		Run: runBatchFork,
@@ -119,7 +153,6 @@ func runBatchFork(ctx *common.RuntimeContext) error {
 	}
 
 	dryRun := parseBool(ctx.Arg("dry-run"))
-
 	summary := batchRepoSummary{
 		Total:   len(repos),
 		Results: make([]batchRepoResult, 0, len(repos)),
@@ -158,6 +191,71 @@ func runBatchFork(ctx *common.RuntimeContext) error {
 	return nil
 }
 
+func newBatchDeleteShortcut() *common.Shortcut {
+	return &common.Shortcut{
+		Name:        "batch-delete",
+		Description: "批量删除仓库，支持逗号分隔列表或 CSV 文件",
+		Flags: []common.Flag{
+			{Name: "repos", Short: "r", Usage: "逗号分隔的仓库标识，格式为 owner/repo，例如: alice/proj1,bob/proj2"},
+			{Name: "from", Usage: "从 CSV 文件读取仓库标识。支持 owner/repo 单列或 owner、repo 双列格式"},
+			{Name: "dry-run", Usage: "仅预览将要删除的仓库，不实际执行", Bool: true, Default: "false"},
+		},
+		Run: runBatchDelete,
+	}
+}
+
+func runBatchDelete(ctx *common.RuntimeContext) error {
+	start := time.Now()
+
+	repos, err := collectRepos(ctx.Arg("repos"), ctx.Arg("from"))
+	if err != nil {
+		return err
+	}
+	if len(repos) == 0 {
+		return fmt.Errorf("未提供仓库标识，请使用 --repos owner/repo1,owner/repo2 或 --from repos.csv")
+	}
+
+	dryRun := parseBool(ctx.Arg("dry-run"))
+	summary := batchRepoSummary{
+		Total:   len(repos),
+		Results: make([]batchRepoResult, 0, len(repos)),
+	}
+
+	for _, repoID := range repos {
+		parts := strings.SplitN(repoID, "/", 2)
+		result := batchRepoResult{Repo: repoID, Action: "delete"}
+		if dryRun {
+			result.Status = "planned"
+			summary.Succeeded++
+			summary.Results = append(summary.Results, result)
+			continue
+		}
+
+		path := fmt.Sprintf("/%s/%s", parts[0], parts[1])
+		if _, err := ctx.CallAPI("DELETE", path, nil); err != nil {
+			result.Status = "failed"
+			result.Error = err.Error()
+			summary.Failed++
+		} else {
+			result.Status = "deleted"
+			summary.Succeeded++
+		}
+		summary.Results = append(summary.Results, result)
+	}
+
+	summary.Duration = time.Since(start).String()
+
+	if err := ctx.OutputData(summary); err != nil {
+		return err
+	}
+	if summary.Failed > 0 {
+		return fmt.Errorf("%d / %d 个仓库删除失败", summary.Failed, summary.Total)
+	}
+	return nil
+}
+
+// --- CSV / list helpers ---
+
 func collectRepos(reposValue, csvPath string) ([]string, error) {
 	repos, err := parseRepoList(reposValue)
 	if err != nil {
@@ -171,7 +269,7 @@ func collectRepos(reposValue, csvPath string) ([]string, error) {
 	if err != nil {
 		return nil, err
 	}
-	return mergeRepoLists(repos, csvRepos), nil
+	return mergeRepoStrings(repos, csvRepos), nil
 }
 
 func parseRepoList(value string) ([]string, error) {
@@ -198,7 +296,6 @@ func readReposFromCSV(path string) ([]string, error) {
 		return nil, nil
 	}
 
-	// Detect column layout: single owner/repo column, or owner+repo dual columns
 	singleCol, ownerCol, repoCol := -1, -1, -1
 	startRow := 0
 	for i, cell := range records[0] {
@@ -210,14 +307,20 @@ func readReposFromCSV(path string) ([]string, error) {
 			ownerCol = i
 			startRow = 1
 		case "repo", "repository", "name":
-			repoCol = i
+			if repoCol == -1 {
+				repoCol = i
+			}
 			startRow = 1
 		}
 	}
 
-	// Fallback: no header — first column is owner/repo
-	if singleCol == -1 && ownerCol == -1 && repoCol == -1 {
-		singleCol = 0
+	if ownerCol == -1 || repoCol == -1 {
+		// Not dual-column: use single-column mode
+		if singleCol == -1 {
+			singleCol = 0
+		}
+		ownerCol = -1
+		repoCol = -1
 	}
 
 	values := make([]string, 0, len(records)-startRow)
@@ -243,10 +346,6 @@ func normalizeRepoIDs(values []string) ([]string, error) {
 		if repoID == "" {
 			continue
 		}
-		parts := strings.SplitN(repoID, "/", 2)
-		if len(parts) != 2 || parts[0] == "" || parts[1] == "" {
-			return nil, fmt.Errorf("无效的仓库标识 %q: 请使用 owner/repo 格式", repoID)
-		}
 		if seen[repoID] {
 			continue
 		}
@@ -256,7 +355,7 @@ func normalizeRepoIDs(values []string) ([]string, error) {
 	return repos, nil
 }
 
-func mergeRepoLists(values ...[]string) []string {
+func mergeRepoStrings(values ...[]string) []string {
 	merged := []string{}
 	seen := map[string]bool{}
 	for _, repos := range values {

shortcuts/repo/repo.go

@@ -12,8 +12,9 @@ import (
 
 func Shortcuts() []*common.Shortcut {
 	return []*common.Shortcut{
-		newBatchDeleteShortcut(),
+		newBatchCreateShortcut(),
 		newBatchForkShortcut(),
+		newBatchDeleteShortcut(),
 		{
 			Name:        "list",
 			Description: "List repositories for a user or organization",
@@ -57,24 +58,9 @@ func Shortcuts() []*common.Shortcut {
 				}
 				targetDir := ctx.Arg("dir")
 
-				var cloneURL string
-				var repoName string
-
-				if strings.HasPrefix(rawURL, "http://") || strings.HasPrefix(rawURL, "https://") {
-					cloneURL = rawURL
-					if !strings.HasSuffix(cloneURL, ".git") {
-						cloneURL += ".git"
-					}
-					parts := strings.Split(strings.TrimSuffix(rawURL, ".git"), "/")
-					repoName = parts[len(parts)-1]
-				} else {
-					parts := strings.SplitN(rawURL, "/", 2)
-					if len(parts) != 2 {
-						return fmt.Errorf("无效的仓库格式 %q: 请使用 owner/repo 格式", rawURL)
-					}
-					webBase := strings.TrimSuffix(strings.TrimSuffix(ctx.Client.BaseURL, "/"), "/api")
-					cloneURL = fmt.Sprintf("%s/%s/%s.git", webBase, parts[0], parts[1])
-					repoName = parts[1]
+				cloneURL, repoName, err := parseCloneTarget(rawURL, ctx.Client.BaseURL)
+				if err != nil {
+					return err
 				}
 
 				if targetDir == "" {
@@ -225,3 +211,30 @@ func Shortcuts() []*common.Shortcut {
 		},
 	}
 }
+
+// parseCloneTarget 把用户输入的 --url 解析成可克隆的 cloneURL 和默认目录名 repoName。
+// 支持两种输入：
+//   - 完整 URL：https://www.gitlink.org.cn/owner/repo[.git]
+//   - 简写格式：owner/repo（会基于 BaseURL 拼成完整 URL）
+//
+// 抽成纯函数便于跨平台单元测试（不依赖 git / PATH）。
+func parseCloneTarget(rawURL, baseURL string) (cloneURL, repoName string, err error) {
+	if strings.HasPrefix(rawURL, "http://") || strings.HasPrefix(rawURL, "https://") {
+		cloneURL = rawURL
+		if !strings.HasSuffix(cloneURL, ".git") {
+			cloneURL += ".git"
+		}
+		parts := strings.Split(strings.TrimSuffix(rawURL, ".git"), "/")
+		repoName = parts[len(parts)-1]
+		return
+	}
+	parts := strings.SplitN(rawURL, "/", 2)
+	if len(parts) != 2 {
+		err = fmt.Errorf("无效的仓库格式 %q: 请使用 owner/repo 格式", rawURL)
+		return
+	}
+	webBase := strings.TrimSuffix(strings.TrimSuffix(baseURL, "/"), "/api")
+	cloneURL = fmt.Sprintf("%s/%s/%s.git", webBase, parts[0], parts[1])
+	repoName = parts[1]
+	return
+}

shortcuts/repo/repo_test.go

@@ -5,8 +5,6 @@ import (
 	"fmt"
 	"net/http"
 	"net/http/httptest"
-	"os"
-	"path/filepath"
 	"strings"
 	"testing"
 
@@ -219,50 +217,38 @@ func assertEqual(t *testing.T, got interface{}, want interface{}) {
 	}
 }
 
-func TestRepoCloneParsesOwnerRepoFormat(t *testing.T) {
-	tmpDir := t.TempDir()
-	fakeGit := createFakeGit(t, tmpDir)
+// clone 测试直接验证 parseCloneTarget 的解析逻辑（纯函数，不依赖 git/PATH，全平台可跑）。
+const testCloneBaseURL = "https://www.gitlink.org.cn/api"
 
-	err := runRepoCloneShortcut(t, fakeGit, map[string]string{
-		"url": "myorg/myrepo",
-	})
+func TestRepoCloneParsesOwnerRepoFormat(t *testing.T) {
+	cloneURL, repoName, err := parseCloneTarget("myorg/myrepo", testCloneBaseURL)
 	if err != nil {
-		t.Fatalf("clone shortcut failed: %v", err)
+		t.Fatalf("parseCloneTarget failed: %v", err)
+	}
+	if cloneURL != "https://www.gitlink.org.cn/myorg/myrepo.git" {
+		t.Errorf("cloneURL = %q, want .../myorg/myrepo.git", cloneURL)
+	}
+	if repoName != "myrepo" {
+		t.Errorf("repoName = %q, want myrepo", repoName)
 	}
 }
 
 func TestRepoCloneParsesFullURL(t *testing.T) {
-	tmpDir := t.TempDir()
-	fakeGit := createFakeGit(t, tmpDir)
-
-	err := runRepoCloneShortcut(t, fakeGit, map[string]string{
-		"url": "https://www.gitlink.org.cn/myorg/myrepo",
-	})
+	cloneURL, repoName, err := parseCloneTarget("https://www.gitlink.org.cn/myorg/myrepo", testCloneBaseURL)
 	if err != nil {
-		t.Fatalf("clone shortcut failed: %v", err)
+		t.Fatalf("parseCloneTarget failed: %v", err)
 	}
-}
-
-func TestRepoCloneCustomTargetDir(t *testing.T) {
-	tmpDir := t.TempDir()
-	fakeGit := createFakeGit(t, tmpDir)
-
-	err := runRepoCloneShortcut(t, fakeGit, map[string]string{
-		"url": "myorg/myrepo",
-		"dir": "/custom/dir",
-	})
-	if err != nil {
-		t.Fatalf("clone shortcut failed: %v", err)
+	// 完整 URL 缺 .git 时应自动补上
+	if cloneURL != "https://www.gitlink.org.cn/myorg/myrepo.git" {
+		t.Errorf("cloneURL = %q, want .../myorg/myrepo.git", cloneURL)
+	}
+	if repoName != "myrepo" {
+		t.Errorf("repoName = %q, want myrepo", repoName)
 	}
 }
 
 func TestRepoCloneInvalidFormat(t *testing.T) {
-	tmpDir := t.TempDir()
-	fakeGit := createFakeGit(t, tmpDir)
-
-	err := runRepoCloneShortcut(t, fakeGit, map[string]string{
-		"url": "invalidformat",
-	})
+	_, _, err := parseCloneTarget("invalidformat", testCloneBaseURL)
 	if err == nil {
 		t.Fatal("expected error for invalid format, got nil")
 	}
@@ -272,52 +258,17 @@ func TestRepoCloneInvalidFormat(t *testing.T) {
 }
 
 func TestRepoCloneURLWithGitSuffix(t *testing.T) {
-	tmpDir := t.TempDir()
-	fakeGit := createFakeGit(t, tmpDir)
-
-	err := runRepoCloneShortcut(t, fakeGit, map[string]string{
-		"url": "https://www.gitlink.org.cn/myorg/myrepo.git",
-	})
+	cloneURL, repoName, err := parseCloneTarget("https://www.gitlink.org.cn/myorg/myrepo.git", testCloneBaseURL)
 	if err != nil {
-		t.Fatalf("clone shortcut failed: %v", err)
+		t.Fatalf("parseCloneTarget failed: %v", err)
 	}
-}
-
-func createFakeGit(t *testing.T, dir string) string {
-	t.Helper()
-	fakeGitPath := filepath.Join(dir, "git")
-	script := `#!/bin/bash
-# Record arguments for test verification
-echo "git $@" > ` + filepath.Join(dir, "git_args.txt") + `
-exit 0
-`
-	if err := os.WriteFile(fakeGitPath, []byte(script), 0755); err != nil {
-		t.Fatalf("failed to create fake git: %v", err)
+	// 已带 .git 时不应重复追加
+	if cloneURL != "https://www.gitlink.org.cn/myorg/myrepo.git" {
+		t.Errorf("cloneURL = %q, want unchanged .../myorg/myrepo.git", cloneURL)
 	}
-	return fakeGitPath
-}
-
-func runRepoCloneShortcut(t *testing.T, fakeGitPath string, args map[string]string) error {
-	t.Helper()
-	shortcut := findRepoShortcut(t, "clone")
-
-	// Set up PATH to include fake git
-	origPath := os.Getenv("PATH")
-	gitDir := filepath.Dir(fakeGitPath)
-	os.Setenv("PATH", gitDir+":"+origPath)
-	defer os.Setenv("PATH", origPath)
-
-	ctx := &common.RuntimeContext{
-		Client: &client.Client{
-			HTTP:    nil,
-			BaseURL: "https://www.gitlink.org.cn/api",
-		},
-		Owner:  "owner",
-		Repo:   "repo",
-		Format: "json",
-		Args:   args,
+	if repoName != "myrepo" {
+		t.Errorf("repoName = %q, want myrepo", repoName)
 	}
-	return shortcut.Run(ctx)
 }
 
 func TestRepoListWithoutUser(t *testing.T) {

shortcuts/workflow/ai/client.go

@@ -0,0 +1,195 @@
+package ai
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/internal/config"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+const deepseekBaseURL = "https://api.deepseek.com/v1/chat/completions"
+const defaultModel = "deepseek-chat"
+
+// AIClient wraps the DeepSeek API for skill step execution.
+type AIClient struct {
+	apiKey string
+	model  string
+	http   *http.Client
+}
+
+const jsonOutputInstruction = `
+
+IMPORTANT: You MUST respond with a single JSON object in exactly this format:
+{"analysis": "<your full markdown report as a single string>", "actions": [...list of actions...]}
+
+The "analysis" field is a STRING containing your entire analysis report in markdown.
+The "actions" field is an array of action objects: {"type":"cli", "module":"pr", "command":"+comment", "args":{"id":"4", "body":"..."}} or {"type":"api", "method":"POST", "path":"/v1/...", "body":{...}}.
+Do NOT include any text outside the JSON object. Do NOT use markdown code fences around the JSON.`
+
+// NewAIClient resolves the API key (env → config) and returns a client, or nil if unavailable.
+func NewAIClient() *AIClient {
+	key := os.Getenv("DEEPSEEK_API_KEY")
+	if key == "" {
+		cfg, err := config.Load()
+		if err == nil {
+			key = cfg.DeepSeekAPIKey
+		}
+	}
+	if key == "" {
+		return nil
+	}
+	return &AIClient{
+		apiKey: key,
+		model:  defaultModel,
+		http:   &http.Client{Timeout: 120 * time.Second},
+	}
+}
+
+// Analyze sends the skill prompt + upstream data to the DeepSeek API and parses the response.
+func (c *AIClient) Analyze(req *wf.AIRequest) (*wf.AIResponse, error) {
+	if c == nil {
+		return nil, fmt.Errorf("AI client not configured: set DEEPSEEK_API_KEY or configure deepseek_api_key")
+	}
+
+	body := map[string]interface{}{
+		"model":       c.model,
+		"max_tokens":  4096,
+		"temperature": 0.1,
+		"messages": []map[string]string{
+			{"role": "system", "content": req.SystemPrompt + jsonOutputInstruction},
+			{"role": "user", "content": req.UserData},
+		},
+		"response_format": map[string]string{"type": "json_object"},
+	}
+
+	payload, err := json.Marshal(body)
+	if err != nil {
+		return nil, fmt.Errorf("marshal request: %w", err)
+	}
+
+	httpReq, err := http.NewRequest("POST", deepseekBaseURL, bytes.NewReader(payload))
+	if err != nil {
+		return nil, fmt.Errorf("create request: %w", err)
+	}
+	httpReq.Header.Set("Content-Type", "application/json")
+	httpReq.Header.Set("Authorization", "Bearer "+c.apiKey)
+
+	resp, err := c.http.Do(httpReq)
+	if err != nil {
+		return nil, fmt.Errorf("API call: %w", err)
+	}
+	defer resp.Body.Close()
+
+	respBody, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return nil, fmt.Errorf("read response: %w", err)
+	}
+
+	if resp.StatusCode != 200 {
+		return nil, fmt.Errorf("DeepSeek API returned %d: %s", resp.StatusCode, string(respBody))
+	}
+
+	var result struct {
+		Choices []struct {
+			Message struct {
+				Content string `json:"content"`
+			} `json:"message"`
+		} `json:"choices"`
+	}
+	if err := json.Unmarshal(respBody, &result); err != nil {
+		return nil, fmt.Errorf("parse response: %w", err)
+	}
+
+	if len(result.Choices) == 0 {
+		return nil, fmt.Errorf("empty response from DeepSeek API")
+	}
+
+	text := result.Choices[0].Message.Content
+	aiResp, err := parseAIResponse(text)
+	if err != nil {
+		if extracted := extractJSONFromMarkdown(text); extracted != "" {
+			aiResp2, err2 := parseAIResponse(extracted)
+			if err2 != nil {
+				return nil, fmt.Errorf("parse AI JSON output: %w\nraw: %s", err, text)
+			}
+			aiResp = aiResp2
+		} else {
+			return nil, fmt.Errorf("parse AI JSON output: %w\nraw: %s", err, text)
+		}
+	}
+
+	return aiResp, nil
+}
+
+// parseAIResponse unmarshals the AI's JSON output, coercing numeric arg values to strings.
+func parseAIResponse(text string) (*wf.AIResponse, error) {
+	raw := struct {
+		Analysis json.RawMessage `json:"analysis"`
+		Actions  []struct {
+			Type    string                 `json:"type"`
+			Method  string                 `json:"method,omitempty"`
+			Path    string                 `json:"path,omitempty"`
+			Body    map[string]interface{} `json:"body,omitempty"`
+			Module  string                 `json:"module,omitempty"`
+			Command string                 `json:"command,omitempty"`
+			Args    map[string]interface{} `json:"args,omitempty"`
+		} `json:"actions"`
+	}{}
+	if err := json.Unmarshal([]byte(text), &raw); err != nil {
+		return nil, err
+	}
+
+	resp := &wf.AIResponse{}
+	if err := json.Unmarshal(raw.Analysis, &resp.Analysis); err != nil {
+		resp.Analysis = string(raw.Analysis)
+	}
+	for _, a := range raw.Actions {
+		args := make(map[string]string, len(a.Args))
+		for k, v := range a.Args {
+			args[k] = fmt.Sprint(v)
+		}
+		resp.Actions = append(resp.Actions, wf.AIAction{
+			Type:    a.Type,
+			Method:  a.Method,
+			Path:    a.Path,
+			Body:    a.Body,
+			Module:  a.Module,
+			Command: a.Command,
+			Args:    args,
+		})
+	}
+	return resp, nil
+}
+
+// HasKey reports whether the AI client is configured.
+func (c *AIClient) HasKey() bool {
+	return c != nil && c.apiKey != ""
+}
+
+// extractJSONFromMarkdown tries to pull a JSON object out of a markdown code fence.
+func extractJSONFromMarkdown(text string) string {
+	start := strings.Index(text, "```json")
+	if start == -1 {
+		start = strings.Index(text, "```")
+	}
+	if start == -1 {
+		return ""
+	}
+	nl := strings.Index(text[start:], "\n")
+	if nl == -1 {
+		return ""
+	}
+	content := text[start+nl+1:]
+	end := strings.Index(content, "```")
+	if end == -1 {
+		return ""
+	}
+	return strings.TrimSpace(content[:end])
+}

shortcuts/workflow/cli/commands.go

@@ -0,0 +1,397 @@
+package cli
+
+import (
+	"fmt"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/internal/output"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/daemon"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/engine"
+)
+
+// Shortcuts returns all workflow CLI commands.
+func Shortcuts() []*common.Shortcut {
+	return []*common.Shortcut{
+		{
+			Name:        "list",
+			Description: "List available workflows",
+			Flags: []common.Flag{
+				{Name: "category", Short: "c", Usage: "Filter by category", Default: ""},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				cat := ctx.Arg("category")
+				workflows := wf.All()
+				filtered := make([]*wf.WorkflowDef, 0)
+				for _, w := range workflows {
+					if cat == "" || strings.EqualFold(w.Category, cat) {
+						filtered = append(filtered, w)
+					}
+				}
+				type listItem struct {
+					Name        string `json:"name"`
+					Category    string `json:"category"`
+					Description string `json:"description"`
+					StepCount   int    `json:"step_count"`
+				}
+				items := make([]listItem, len(filtered))
+				for i, w := range filtered {
+					items[i] = listItem{
+						Name:        w.Name,
+						Category:    w.Category,
+						Description: w.Description,
+						StepCount:   len(w.Steps),
+					}
+				}
+				return ctx.OutputData(items)
+			},
+		},
+		{
+			Name:        "info",
+			Description: "Show workflow detail (steps and trigger)",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				w := wf.Get(name)
+				if w == nil {
+					return fmt.Errorf("workflow %q not found; use workflow +list to see available workflows", name)
+				}
+				return ctx.OutputData(w)
+			},
+		},
+		{
+			Name:        "run",
+			Description: "Execute a workflow manually",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+				{Name: "dry-run", Usage: "Preview mode (no AI calls)", Bool: true},
+				{Name: "ai", Usage: "Force AI mode (requires API key)", Bool: true},
+				{Name: "no-ai", Usage: "Force rule engine mode (no AI)", Bool: true},
+				{Name: "desc", Short: "d", Usage: "Project description (for project-init workflow)", Default: ""},
+				{Name: "repos", Usage: "Comma-separated repositories for multi-repo workflow, e.g. org/backend,org/frontend", Default: ""},
+				{Name: "from", Usage: "CSV file for multi-repo workflow with owner,repo columns", Default: ""},
+				{Name: "release", Usage: "Target release/tag for multi-repo release coordination", Default: ""},
+				{Name: "wiki-repo", Usage: "Wiki target repo for publishing multi-repo report, e.g. org/dashboard", Default: ""},
+				{Name: "daemon-loop", Usage: "Internal: run in loop mode", Bool: true},
+				{Name: "interval", Usage: "Internal: loop interval", Default: "5m"},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				w := wf.Get(name)
+				if w == nil {
+					return fmt.Errorf("workflow %q not found; use workflow +list to see available workflows", name)
+				}
+
+				if desc := ctx.Arg("desc"); desc != "" {
+					ctx.Args["_desc"] = desc
+				}
+
+				if wr := ctx.Arg("wiki-repo"); wr != "" {
+					owner, repo := splitRepoRef(wr)
+					if owner != "" && repo != "" {
+						ctx.Args["_wiki_owner"] = owner
+						ctx.Args["_wiki_repo"] = repo
+					}
+				}
+
+				aiMode, err := resolveAIModeFromArgs(ctx)
+				if err != nil {
+					return err
+				}
+
+				if ctx.Arg("daemon-loop") == "true" {
+					intervalStr := ctx.Arg("interval")
+					interval, err := time.ParseDuration(intervalStr)
+					if err != nil {
+						return fmt.Errorf("invalid interval %q: %w", intervalStr, err)
+					}
+					return daemon.DaemonLoop(ctx, w, interval)
+				}
+
+				return runWorkflowCommand(ctx, w, ctx.Arg("dry-run") == "true", aiMode)
+			},
+		},
+		{
+			Name:        "watch",
+			Description: "Poll for changes and trigger workflow on delta",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+				{Name: "interval", Short: "i", Usage: "Poll interval (e.g. 30s, 5m, 1h)", Default: "5m"},
+				{Name: "step", Short: "s", Usage: "Step name to watch for changes (default: first step)", Default: ""},
+				{Name: "ai", Usage: "Force AI mode (requires API key)", Bool: true},
+				{Name: "no-ai", Usage: "Force rule engine mode (no AI)", Bool: true},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				w := wf.Get(name)
+				if w == nil {
+					return fmt.Errorf("workflow %q not found; use workflow +list to see available workflows", name)
+				}
+				if w.Name == "multi-repo" {
+					return fmt.Errorf("workflow +watch 不支持 multi-repo；多仓库协同请求量较大，请使用 workflow +run 手动检查，或 workflow +schedule --interval 6h/24h 做低频巡检")
+				}
+				intervalStr := ctx.Arg("interval")
+				interval, err := time.ParseDuration(intervalStr)
+				if err != nil {
+					return fmt.Errorf("invalid interval %q: %w", intervalStr, err)
+				}
+
+				aiMode, modeErr := resolveAIModeFromArgs(ctx)
+				if modeErr != nil {
+					return modeErr
+				}
+				ctx.AIMode = aiMode
+
+				return daemon.Watch(ctx, w, interval, ctx.Arg("step"))
+			},
+		},
+		{
+			Name:        "schedule",
+			Description: "Run a workflow on a repeating schedule",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+				{Name: "interval", Short: "i", Usage: "Run interval (e.g. 1h, 24h)", Default: "24h"},
+				{Name: "ai", Usage: "Force AI mode (requires API key)", Bool: true},
+				{Name: "no-ai", Usage: "Force rule engine mode (no AI)", Bool: true},
+				{Name: "repos", Usage: "Comma-separated repositories for multi-repo workflow", Default: ""},
+				{Name: "from", Usage: "CSV file for multi-repo workflow with owner,repo columns", Default: ""},
+				{Name: "release", Usage: "Target release/tag for multi-repo release coordination", Default: ""},
+				{Name: "wiki-repo", Usage: "Wiki target repo for publishing multi-repo report, e.g. org/dashboard", Default: ""},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				w := wf.Get(name)
+				if w == nil {
+					return fmt.Errorf("workflow %q not found; use workflow +list to see available workflows", name)
+				}
+				intervalStr := ctx.Arg("interval")
+				interval, err := time.ParseDuration(intervalStr)
+				if err != nil {
+					return fmt.Errorf("invalid interval %q: %w", intervalStr, err)
+				}
+
+				if wr := ctx.Arg("wiki-repo"); wr != "" {
+					owner, repo := splitRepoRef(wr)
+					if owner != "" && repo != "" {
+						ctx.Args["_wiki_owner"] = owner
+						ctx.Args["_wiki_repo"] = repo
+					}
+				}
+
+				aiMode, modeErr := resolveAIModeFromArgs(ctx)
+				if modeErr != nil {
+					return modeErr
+				}
+				ctx.AIMode = aiMode
+
+				return daemon.Schedule(ctx, w, interval)
+			},
+		},
+		{
+			Name:        "start",
+			Description: "Start workflow as background daemon",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+				{Name: "interval", Short: "i", Usage: "Poll interval (e.g. 5m, 1h)", Default: "5m"},
+				{Name: "ai", Usage: "Force AI mode (requires API key)", Bool: true},
+				{Name: "no-ai", Usage: "Force rule engine mode (no AI)", Bool: true},
+				{Name: "repos", Usage: "Comma-separated repositories for multi-repo workflow", Default: ""},
+				{Name: "from", Usage: "CSV file for multi-repo workflow with owner,repo columns", Default: ""},
+				{Name: "release", Usage: "Target release/tag for multi-repo release coordination", Default: ""},
+				{Name: "wiki-repo", Usage: "Wiki target repo for publishing multi-repo report, e.g. org/dashboard", Default: ""},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				w := wf.Get(name)
+				if w == nil {
+					return fmt.Errorf("workflow %q not found; use workflow +list to see available workflows", name)
+				}
+				intervalStr := ctx.Arg("interval")
+				interval, err := time.ParseDuration(intervalStr)
+				if err != nil {
+					return fmt.Errorf("invalid interval %q: %w", intervalStr, err)
+				}
+
+				if wr := ctx.Arg("wiki-repo"); wr != "" {
+					owner, repo := splitRepoRef(wr)
+					if owner != "" && repo != "" {
+						ctx.Args["_wiki_owner"] = owner
+						ctx.Args["_wiki_repo"] = repo
+					}
+				}
+
+				aiMode, modeErr := resolveAIModeFromArgs(ctx)
+				if modeErr != nil {
+					return modeErr
+				}
+
+				return daemon.StartDaemon(ctx, w, interval, aiMode)
+			},
+		},
+		{
+			Name:        "stop",
+			Description: "Stop workflow daemon",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				return daemon.StopDaemon(name)
+			},
+		},
+		{
+			Name:        "status",
+			Description: "Show daemon status",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				return daemon.StatusDaemon(name)
+			},
+		},
+		{
+			Name:        "logs",
+			Description: "View daemon log output",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+				{Name: "follow", Short: "f", Usage: "Follow log output (like tail -f)", Bool: true},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				return daemon.TailDaemonLog(name, ctx.Arg("follow") == "true")
+			},
+		},
+		{
+			Name:        "install-systemd",
+			Description: "Generate systemd service unit for a workflow daemon",
+			Flags: []common.Flag{
+				{Name: "name", Short: "n", Usage: "Workflow name", Required: true},
+				{Name: "interval", Short: "i", Usage: "Poll interval", Default: "5m"},
+				{Name: "ai", Usage: "Force AI mode (requires API key)", Bool: true},
+				{Name: "no-ai", Usage: "Force rule engine mode (no AI)", Bool: true},
+			},
+			Run: func(ctx *common.RuntimeContext) error {
+				name, err := ctx.RequireArg("name")
+				if err != nil {
+					return err
+				}
+				w := wf.Get(name)
+				if w == nil {
+					return fmt.Errorf("workflow %q not found; use workflow +list to see available workflows", name)
+				}
+
+				aiMode, modeErr := resolveAIModeFromArgs(ctx)
+				if modeErr != nil {
+					return modeErr
+				}
+
+				return daemon.InstallSystemdUnit(ctx, w, ctx.Arg("interval"), aiMode)
+			},
+		},
+	}
+}
+
+func resolveAIModeFromArgs(ctx *common.RuntimeContext) (string, error) {
+	ai := ctx.Arg("ai") == "true"
+	noAI := ctx.Arg("no-ai") == "true"
+	if ai && noAI {
+		return "", fmt.Errorf("--ai 和 --no-ai 互斥，只能指定其中一个")
+	}
+	if ai {
+		return "ai", nil
+	}
+	if noAI {
+		return "no-ai", nil
+	}
+	return "auto", nil
+}
+
+func runWorkflowCommand(ctx *common.RuntimeContext, w *wf.WorkflowDef, dryRun bool, aiMode string) error {
+	result, err := engine.RunWithMode(ctx, w, dryRun, aiMode)
+	if err != nil {
+		return err
+	}
+
+	needsAI := 0
+	ruleEngine := 0
+	for _, sr := range result.Steps {
+		if m, ok := sr.Data.(map[string]interface{}); ok {
+			if v, _ := m["_needs_ai"]; v == true {
+				needsAI++
+			}
+			if v, _ := m["_ai_used"]; v == true {
+				ruleEngine++
+			}
+		}
+	}
+
+	if needsAI > 0 {
+		fmt.Fprintf(os.Stderr, "\n⚠ %d 个 skill 步骤需要 AI 处理:\n", needsAI)
+		for _, sr := range result.Steps {
+			if m, ok := sr.Data.(map[string]interface{}); ok {
+				if v, _ := m["_needs_ai"]; v == true {
+					fmt.Fprintf(os.Stderr, "   - %s (%s)\n", sr.Step, m["_skill"])
+				}
+			}
+		}
+		fmt.Fprintf(os.Stderr, "\n你可以:\n")
+		fmt.Fprintf(os.Stderr, "  1. 配置 API Key 启用全自动: gitlink-cli config set deepseek_api_key <key>\n")
+		fmt.Fprintf(os.Stderr, "  2. 将以上完整 JSON 输出交给 AI Agent 继续处理\n")
+	} else if ruleEngine > 0 {
+		fmt.Fprintf(os.Stderr, "🤖 AI 已处理 %d 个 skill 步骤\n", ruleEngine)
+	} else {
+		noAI := 0
+		for _, sr := range result.Steps {
+			if m, ok := sr.Data.(map[string]interface{}); ok {
+				if v, _ := m["_ai_used"]; v == false {
+					if _, hasSkill := m["_skill"]; hasSkill {
+						noAI++
+					}
+				}
+			}
+		}
+		if noAI > 0 {
+			fmt.Fprintf(os.Stderr, "⚙️  规则引擎已处理 %d 个 skill 步骤 (未使用 AI)\n", noAI)
+		}
+	}
+
+	return ctx.Output(output.SuccessEnvelope(result, nil))
+}
+
+func splitRepoRef(raw string) (owner, repo string) {
+	parts := strings.SplitN(strings.TrimSpace(raw), "/", 2)
+	if len(parts) == 2 {
+		return strings.TrimSpace(parts[0]), strings.TrimSpace(parts[1])
+	}
+	return "", ""
+}

shortcuts/workflow/cli/commands_test.go

@@ -0,0 +1,26 @@
+package cli
+
+import "testing"
+
+func TestShortcutsCount(t *testing.T) {
+	sc := Shortcuts()
+	if len(sc) != 10 {
+		t.Fatalf("expected 10 shortcuts (list, info, run, watch, schedule, start, stop, status, logs, install-systemd), got %d", len(sc))
+	}
+	names := map[string]bool{
+		"list": false, "info": false, "run": false, "watch": false,
+		"schedule": false, "start": false, "stop": false, "status": false,
+		"logs": false, "install-systemd": false,
+	}
+	for _, s := range sc {
+		if _, ok := names[s.Name]; !ok {
+			t.Fatalf("unexpected shortcut: %s", s.Name)
+		}
+		names[s.Name] = true
+	}
+	for n, found := range names {
+		if !found {
+			t.Fatalf("missing shortcut: %s", n)
+		}
+	}
+}

shortcuts/workflow/command_util.go

@@ -0,0 +1,68 @@
+package workflow
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+)
+
+// ParseCommandTarget splits a CLI command string into tokens, respecting quoted arguments.
+func ParseCommandTarget(target string) []string {
+	var parts []string
+	var current strings.Builder
+	inQuote := false
+	quoteChar := byte(0)
+
+	for i := 0; i < len(target); i++ {
+		c := target[i]
+		switch {
+		case c == '"' || c == '\'':
+			if inQuote && c == quoteChar {
+				inQuote = false
+				quoteChar = 0
+			} else if !inQuote {
+				inQuote = true
+				quoteChar = c
+			} else {
+				current.WriteByte(c)
+			}
+		case c == ' ' && !inQuote:
+			if current.Len() > 0 {
+				parts = append(parts, current.String())
+				current.Reset()
+			}
+		default:
+			current.WriteByte(c)
+		}
+	}
+	if current.Len() > 0 {
+		parts = append(parts, current.String())
+	}
+	return parts
+}
+
+// ResolveCLIBinary finds the gitlink-cli binary for subprocess calls.
+func ResolveCLIBinary() string {
+	if exe, err := os.Executable(); err == nil && exe != "" {
+		return exe
+	}
+	for _, p := range []string{"./gitlink-cli", "./gitlink-cli.exe", "../gitlink-cli", "../gitlink-cli.exe"} {
+		if _, err := os.Stat(p); err == nil {
+			if abs, err := filepath.Abs(p); err == nil {
+				return abs
+			}
+			return p
+		}
+	}
+	return "gitlink-cli"
+}
+
+// ResolvePath replaces template placeholders in a path string.
+func ResolvePath(template, owner, repo string) string {
+	base := fmt.Sprintf("/%s/%s", owner, repo)
+	v1 := fmt.Sprintf("/v1/%s/%s", owner, repo)
+	s := strings.Replace(template, "{v1}", v1, 1)
+	s = strings.Replace(s, "{base}", base, 1)
+	return s
+}

shortcuts/workflow/daemon/daemon.go

@@ -0,0 +1,397 @@
+package daemon
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"os/exec"
+	"os/signal"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"syscall"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/internal/config"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/engine"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/state"
+)
+
+// StartDaemon launches a workflow as a background daemon process.
+func StartDaemon(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, interval time.Duration, aiMode string) error {
+	bin, err := os.Executable()
+	if err != nil {
+		return fmt.Errorf("cannot find executable: %w", err)
+	}
+
+	args := BuildDaemonArgs(ctx, wfDef, interval, aiMode)
+
+	cmd := exec.Command(bin, args...)
+	applyDaemonAttrs(cmd)
+
+	logPath := daemonLogPath(wfDef.Name)
+	logFile, err := os.OpenFile(logPath, os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0600)
+	if err != nil {
+		return fmt.Errorf("create log file: %w", err)
+	}
+	cmd.Stdout = logFile
+	cmd.Stderr = logFile
+	cmd.Stdin = nil
+
+	if err := cmd.Start(); err != nil {
+		logFile.Close()
+		return fmt.Errorf("start daemon: %w", err)
+	}
+
+	pid := cmd.Process.Pid
+	if err := savePID(wfDef.Name, pid); err != nil {
+		return fmt.Errorf("save pid: %w", err)
+	}
+
+	fmt.Printf("Daemon started for %q (PID: %d)\n", wfDef.Name, pid)
+	fmt.Printf("Log: %s\n", logPath)
+	return nil
+}
+
+// BuildDaemonArgs constructs the CLI arguments for the daemon subprocess.
+func BuildDaemonArgs(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, interval time.Duration, aiMode string) []string {
+	args := []string{
+		"workflow", "+run", "--name", wfDef.Name,
+		"--format", "json", "--daemon-loop",
+		"--interval", interval.String(),
+	}
+	if !engine.IsExplicitMultiRepoRun(ctx, wfDef) {
+		args = append(args, "--owner", ctx.Owner, "--repo", ctx.Repo)
+	}
+	if aiMode == "ai" {
+		args = append(args, "--ai")
+	} else if aiMode == "no-ai" {
+		args = append(args, "--no-ai")
+	}
+	if repos := ctx.Arg("repos"); repos != "" {
+		args = append(args, "--repos", repos)
+	}
+	if from := ctx.Arg("from"); from != "" {
+		args = append(args, "--from", from)
+	}
+	if release := ctx.Arg("release"); release != "" {
+		args = append(args, "--release", release)
+	}
+	return args
+}
+
+// StopDaemon stops a running workflow daemon by name.
+func StopDaemon(name string) error {
+	pid, err := readPID(name)
+	if err != nil {
+		return err
+	}
+
+	proc, err := os.FindProcess(pid)
+	if err != nil {
+		cleanPID(name)
+		return fmt.Errorf("daemon %q not running (PID %d not found)", name, pid)
+	}
+
+	if err := proc.Signal(os.Interrupt); err != nil {
+		cleanPID(name)
+		return fmt.Errorf("failed to stop daemon %q: %w", name, err)
+	}
+
+	cleanPID(name)
+	fmt.Printf("Daemon %q stopped (PID %d)\n", name, pid)
+	return nil
+}
+
+// StatusDaemon prints the current daemon status for a workflow.
+func StatusDaemon(name string) error {
+	st, err := state.LoadState(name)
+	if err != nil {
+		return fmt.Errorf("load state: %w", err)
+	}
+
+	pid, pidErr := readPID(name)
+	running := pidErr == nil && processRunning(pid)
+
+	fmt.Printf("工作流: %s\n", name)
+	if running {
+		fmt.Printf("状态: 运行中 (PID: %d)\n", pid)
+	} else {
+		fmt.Println("状态: 已停止")
+	}
+	if st.LastRun != "" {
+		t, err := time.Parse(time.RFC3339, st.LastRun)
+		if err == nil {
+			fmt.Printf("上次运行: %s\n", t.Format("2006-01-02 15:04"))
+		} else {
+			fmt.Printf("上次运行: %s\n", st.LastRun)
+		}
+	}
+	fmt.Printf("累计运行: %d 次\n", st.TotalRuns)
+	fmt.Printf("快照步骤: %d 个\n", len(st.Snapshots))
+	fmt.Printf("日志文件: %s\n", daemonLogPath(name))
+	return nil
+}
+
+// DaemonLoop runs the workflow repeatedly in a loop.
+func DaemonLoop(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, interval time.Duration) error {
+	tick := time.NewTicker(interval)
+	defer tick.Stop()
+
+	doDaemonCycle(ctx, wfDef)
+
+	for {
+		select {
+		case <-tick.C:
+			doDaemonCycle(ctx, wfDef)
+		}
+	}
+}
+
+func doDaemonCycle(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef) {
+	st, _ := state.LoadState(wfDef.Name)
+
+	dryResult, err := engine.Run(ctx, wfDef, true)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "[%s] error: %v\n", time.Now().Format(time.RFC3339), err)
+		return
+	}
+
+	changed := st.Diff(dryResult.Steps)
+	if len(changed) == 0 {
+		if st.TotalRuns > 0 {
+			fmt.Fprintf(os.Stderr, "[%s] 没有检测到变更\n", time.Now().Format(time.RFC3339))
+			st.TotalRuns++
+			st.Save()
+			return
+		}
+	} else {
+		fmt.Fprintf(os.Stderr, "[%s] 🔔 检测到变更: %v\n", time.Now().Format(time.RFC3339), changed)
+	}
+
+	fullResult, err := engine.Run(ctx, wfDef, false)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "[%s] run error: %v\n", time.Now().Format(time.RFC3339), err)
+		return
+	}
+
+	st, _ = state.LoadState(wfDef.Name)
+	st.TotalRuns++
+	st.Diff(fullResult.Steps)
+	st.UpdateSnapshots(fullResult.Steps)
+	st.Save()
+
+	ok, total := 0, len(fullResult.Steps)
+	for _, sr := range fullResult.Steps {
+		if sr.OK {
+			ok++
+		}
+	}
+	fmt.Fprintf(os.Stderr, "[%s] ✅ %d/%d steps OK\n", time.Now().Format(time.RFC3339), ok, total)
+
+	for _, sr := range fullResult.Steps {
+		if !sr.OK {
+			fmt.Fprintf(os.Stderr, "[%s] ❌ %s 失败: %s\n", time.Now().Format(time.RFC3339), sr.Step, sr.Error)
+		}
+		if sr.Type == wf.StepTypeSkill && sr.Data != nil {
+			logSkillFindings(sr)
+		}
+	}
+}
+
+func logSkillFindings(sr wf.StepResult) {
+	m, ok := sr.Data.(map[string]interface{})
+	if !ok {
+		return
+	}
+	skill, _ := m["_skill"].(string)
+	analysis := m["analysis"]
+
+	fmt.Fprintf(os.Stderr, "[%s] ── %s 分析结果 ──\n", time.Now().Format(time.RFC3339), skill)
+
+	if s, ok := analysis.(string); ok && s != "" {
+		for _, line := range strings.Split(s, "\n") {
+			fmt.Fprintf(os.Stderr, "[%s] %s\n", time.Now().Format(time.RFC3339), line)
+		}
+		return
+	}
+
+	am, _ := analysis.(map[string]interface{})
+	if am == nil {
+		return
+	}
+
+	if summary, ok := am["summary"].(string); ok && summary != "" {
+		fmt.Fprintf(os.Stderr, "[%s] 📋 %s\n", time.Now().Format(time.RFC3339), summary)
+	}
+
+	if findings, ok := am["findings"]; ok && findings != nil {
+		raw, _ := json.Marshal(findings)
+		var arr []interface{}
+		if json.Unmarshal(raw, &arr) == nil {
+			for _, f := range arr {
+				if fm, ok := f.(map[string]interface{}); ok {
+					sev := fm["severity"]
+					what := fm["what"]
+					fmt.Fprintf(os.Stderr, "[%s]   [%v] %v\n", time.Now().Format(time.RFC3339), sev, what)
+					if prNum, ok := fm["pr_number"]; ok && prNum != nil {
+						fmt.Fprintf(os.Stderr, "[%s]     PR: #%v\n", time.Now().Format(time.RFC3339), prNum)
+					}
+				}
+			}
+		}
+	}
+
+	if msg, ok := am["message"].(string); ok && msg != "" {
+		fmt.Fprintf(os.Stderr, "[%s] ℹ️  %s\n", time.Now().Format(time.RFC3339), msg)
+	}
+
+	if reviewData, ok := am["reviewed_prs"]; ok {
+		fmt.Fprintf(os.Stderr, "[%s]   已审查 PR 数: %v\n", time.Now().Format(time.RFC3339), reviewData)
+	}
+	if totalFindings, ok := am["total_findings"]; ok {
+		fmt.Fprintf(os.Stderr, "[%s]   发现问题数: %v\n", time.Now().Format(time.RFC3339), totalFindings)
+	}
+}
+
+// DaemonLogPath returns the log file path for a workflow daemon.
+func DaemonLogPath(name string) string {
+	return daemonLogPath(name)
+}
+
+func daemonLogPath(name string) string {
+	return filepath.Join(config.ConfigDir(), fmt.Sprintf("workflow-%s.log", name))
+}
+
+// TailDaemonLog reads and optionally follows a daemon log file.
+func TailDaemonLog(name string, follow bool) error {
+	path := daemonLogPath(name)
+	data, err := os.ReadFile(path)
+	if err != nil {
+		return fmt.Errorf("读取日志文件 %s: %w (daemon 可能尚未启动)", path, err)
+	}
+	fmt.Print(string(data))
+
+	if !follow {
+		return nil
+	}
+
+	sig := make(chan os.Signal, 1)
+	signal.Notify(sig, os.Interrupt)
+	ticker := time.NewTicker(1 * time.Second)
+	defer ticker.Stop()
+
+	offset := int64(len(data))
+	for {
+		select {
+		case <-sig:
+			return nil
+		case <-ticker.C:
+			fi, err := os.Stat(path)
+			if err != nil {
+				continue
+			}
+			if fi.Size() > offset {
+				f, err := os.Open(path)
+				if err != nil {
+					continue
+				}
+				f.Seek(offset, 0)
+				buf := make([]byte, fi.Size()-offset)
+				n, _ := f.Read(buf)
+				if n > 0 {
+					fmt.Print(string(buf[:n]))
+				}
+				offset = fi.Size()
+				f.Close()
+			}
+		}
+	}
+}
+
+// InstallSystemdUnit generates a systemd service unit file for a workflow daemon.
+func InstallSystemdUnit(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, interval, aiMode string) error {
+	if err := ctx.ResolveOwnerRepo(); err != nil {
+		return fmt.Errorf("resolve repo: %w", err)
+	}
+
+	bin, _ := os.Executable()
+	extraArgs := ""
+	if aiMode == "ai" {
+		extraArgs = " --ai"
+	} else if aiMode == "no-ai" {
+		extraArgs = " --no-ai"
+	}
+
+	unit := fmt.Sprintf(`[Unit]
+Description=GitLink CLI Workflow: %s (%s/%s)
+After=network.target
+
+[Service]
+Type=simple
+ExecStart=%s workflow +run --name %s --owner %s --repo %s --format json --daemon-loop --interval %s%s
+Restart=on-failure
+RestartSec=30
+StandardOutput=append:%s
+StandardError=append:%s
+
+[Install]
+WantedBy=multi-user.target
+`,
+		wfDef.Name, ctx.Owner, ctx.Repo,
+		bin, wfDef.Name, ctx.Owner, ctx.Repo, interval, extraArgs,
+		daemonLogPath(wfDef.Name), daemonLogPath(wfDef.Name),
+	)
+
+	unitPath := filepath.Join(config.ConfigDir(), fmt.Sprintf("workflow-%s.service", wfDef.Name))
+	if err := os.WriteFile(unitPath, []byte(unit), 0644); err != nil {
+		return fmt.Errorf("写入 unit 文件: %w", err)
+	}
+
+	fmt.Printf("Systemd unit 已写入: %s\n\n", unitPath)
+	fmt.Println("安装步骤:")
+	fmt.Printf("  sudo cp %s /etc/systemd/system/\n", unitPath)
+	fmt.Println("  sudo systemctl daemon-reload")
+	fmt.Printf("  sudo systemctl enable workflow-%s\n", wfDef.Name)
+	fmt.Printf("  sudo systemctl start workflow-%s\n", wfDef.Name)
+	fmt.Println()
+	fmt.Printf("查看日志: journalctl -u workflow-%s -f\n", wfDef.Name)
+	return nil
+}
+
+func savePID(name string, pid int) error {
+	dir := config.ConfigDir()
+	if err := os.MkdirAll(dir, 0700); err != nil {
+		return err
+	}
+	return os.WriteFile(pidPath(name), []byte(strconv.Itoa(pid)), 0600)
+}
+
+func readPID(name string) (int, error) {
+	data, err := os.ReadFile(pidPath(name))
+	if err != nil {
+		if os.IsNotExist(err) {
+			return 0, fmt.Errorf("daemon %q is not running (no PID file)", name)
+		}
+		return 0, err
+	}
+	return strconv.Atoi(string(data))
+}
+
+func cleanPID(name string) {
+	os.Remove(pidPath(name))
+}
+
+func processRunning(pid int) bool {
+	proc, err := os.FindProcess(pid)
+	if err != nil {
+		return false
+	}
+	return proc.Signal(syscall.Signal(0)) == nil
+}
+
+func pidPath(name string) string {
+	return filepath.Join(config.ConfigDir(), fmt.Sprintf("workflow-%s.pid", name))
+}

shortcuts/workflow/daemon/daemon_test.go

@@ -0,0 +1,43 @@
+package daemon
+
+import (
+	"testing"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestBuildDaemonArgsSkipsOwnerRepoForExplicitMultiRepo(t *testing.T) {
+	wfDef := &wf.WorkflowDef{
+		Name: "multi-repo",
+	}
+	ctx := &common.RuntimeContext{
+		Owner: "ignored",
+		Repo:  "ignored",
+		Args: map[string]string{
+			"repos":   "org/backend,org/frontend",
+			"release": "v1.4.0",
+		},
+	}
+
+	args := BuildDaemonArgs(ctx, wfDef, 24*time.Hour, "no-ai")
+	if stringSliceContains(args, "--owner") || stringSliceContains(args, "--repo") {
+		t.Fatalf("explicit multi-repo daemon args should not include owner/repo: %v", args)
+	}
+	if !stringSliceContains(args, "--repos") || !stringSliceContains(args, "org/backend,org/frontend") {
+		t.Fatalf("daemon args missing repos: %v", args)
+	}
+	if !stringSliceContains(args, "--release") || !stringSliceContains(args, "v1.4.0") {
+		t.Fatalf("daemon args missing release: %v", args)
+	}
+}
+
+func stringSliceContains(items []string, want string) bool {
+	for _, item := range items {
+		if item == want {
+			return true
+		}
+	}
+	return false
+}

shortcuts/workflow/daemon/proc_unix.go

@@ -0,0 +1,12 @@
+//go:build !windows
+
+package daemon
+
+import (
+	"os/exec"
+	"syscall"
+)
+
+func applyDaemonAttrs(cmd *exec.Cmd) {
+	cmd.SysProcAttr = &syscall.SysProcAttr{Setsid: true}
+}

shortcuts/workflow/daemon/proc_windows.go

@@ -0,0 +1,12 @@
+//go:build windows
+
+package daemon
+
+import (
+	"os/exec"
+	"syscall"
+)
+
+func applyDaemonAttrs(cmd *exec.Cmd) {
+	cmd.SysProcAttr = &syscall.SysProcAttr{HideWindow: true}
+}

shortcuts/workflow/daemon/schedule.go

@@ -0,0 +1,68 @@
+package daemon
+
+import (
+	"fmt"
+	"os"
+	"os/signal"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/engine"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/state"
+)
+
+// Schedule runs the full workflow on a repeating interval. Blocks until Ctrl+C.
+func Schedule(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, interval time.Duration) error {
+	fmt.Printf("⏰ Scheduled %q every %v on %s/%s\n", wfDef.Name, interval, ctx.Owner, ctx.Repo)
+	fmt.Println("   Press Ctrl+C to stop")
+
+	sig := make(chan os.Signal, 1)
+	signal.Notify(sig, os.Interrupt)
+	tick := time.NewTicker(interval)
+	defer tick.Stop()
+
+	st, _ := state.LoadState(wfDef.Name)
+	dryResult, _ := engine.Run(ctx, wfDef, true)
+	st.UpdateSnapshots(dryResult.Steps)
+	st.TotalRuns++
+	st.Save()
+
+	for {
+		select {
+		case <-sig:
+			fmt.Println("\n👋 schedule stopped")
+			return nil
+		case t := <-tick.C:
+			dryResult, err := engine.Run(ctx, wfDef, true)
+			if err != nil {
+				fmt.Printf("[%s] ❌ error: %v\n", t.Format("15:04:05"), err)
+				continue
+			}
+
+			changed := st.Diff(dryResult.Steps)
+			st.UpdateSnapshots(dryResult.Steps)
+			st.TotalRuns++
+			st.Save()
+
+			if len(changed) == 0 {
+				fmt.Printf("[%s] ✓ no changes, skipped AI run\n", t.Format("15:04:05"))
+				continue
+			}
+
+			fmt.Printf("[%s] ⏳ changes detected, running %q with AI...\n", t.Format("15:04:05"), wfDef.Name)
+			result, err := engine.Run(ctx, wfDef, false)
+			if err != nil {
+				fmt.Printf("❌ error: %v\n", err)
+				continue
+			}
+			ok, total := 0, len(result.Steps)
+			for _, sr := range result.Steps {
+				if sr.OK {
+					ok++
+				}
+			}
+			fmt.Printf("✅ %d/%d steps OK\n", ok, total)
+		}
+	}
+}

shortcuts/workflow/daemon/watch.go

@@ -0,0 +1,72 @@
+package daemon
+
+import (
+	"fmt"
+	"os"
+	"os/signal"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/engine"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/state"
+)
+
+// Watch polls the first step (or watchStep) every interval and triggers the
+// full workflow (with AI) only when data changes. Blocks until Ctrl+C.
+func Watch(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, interval time.Duration, watchStep string) error {
+	if watchStep == "" && len(wfDef.Steps) > 0 {
+		watchStep = wfDef.Steps[0].Name
+	}
+
+	fmt.Printf("👀 Watching %s/%s for %q changes every %v\n", ctx.Owner, ctx.Repo, watchStep, interval)
+	fmt.Printf("   Trigger: %s on %s\n", wfDef.Trigger.Type, wfDef.Trigger.On)
+	fmt.Println("   Press Ctrl+C to stop")
+
+	sig := make(chan os.Signal, 1)
+	signal.Notify(sig, os.Interrupt)
+
+	st, _ := state.LoadState(wfDef.Name)
+	tick := time.NewTicker(interval)
+	defer tick.Stop()
+
+	for {
+		select {
+		case <-sig:
+			fmt.Println("\n👋 watch stopped")
+			return nil
+		case t := <-tick.C:
+			dryResult, err := engine.Run(ctx, wfDef, true)
+			if err != nil {
+				fmt.Printf("[%s] ❌ error: %v\n", t.Format("15:04:05"), err)
+				continue
+			}
+
+			changed := st.Diff(dryResult.Steps)
+			if len(changed) == 0 && st.TotalRuns > 0 {
+				fmt.Printf("[%s] ✓ no changes\n", t.Format("15:04:05"))
+				continue
+			}
+
+			fmt.Printf("[%s] 🔔 change detected: %v\n", t.Format("15:04:05"), changed)
+
+			result, err := engine.Run(ctx, wfDef, false)
+			if err != nil {
+				fmt.Printf("[%s] ❌ AI run error: %v\n", t.Format("15:04:05"), err)
+				continue
+			}
+
+			st.UpdateSnapshots(result.Steps)
+			st.TotalRuns++
+			st.Save()
+
+			for _, sr := range result.Steps {
+				if sr.OK {
+					fmt.Printf("   ✓ %s\n", sr.Step)
+				} else {
+					fmt.Printf("   ✗ %s: %s\n", sr.Step, sr.Error)
+				}
+			}
+		}
+	}
+}

shortcuts/workflow/defs/code_quality.go

@@ -0,0 +1,26 @@
+package defs
+
+import wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+
+// RegisterCodeQuality registers the code-quality workflow definition.
+func RegisterCodeQuality() {
+	wf.Register(&wf.WorkflowDef{
+		Name:        "code-quality",
+		Category:    "质量",
+		Description: "代码质量看门人：PR 提交 → Review → CI 检查 → 结果汇总",
+		Trigger: wf.TriggerDef{
+			Type:     "poll",
+			On:       "pr.opened",
+			Interval: "5m",
+		},
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeCommand, Name: "open-prs", Purpose: "获取开放 PR 列表", Target: "pr +list --state open --limit 20"},
+			{Type: wf.StepTypeCommand, Name: "ci-builds", Purpose: "获取 CI 构建状态", Target: "ci +builds --limit 10"},
+			{Type: wf.StepTypeCommand, Name: "repo-info", Purpose: "获取仓库保护规则配置", Target: "repo +info"},
+			{Type: wf.StepTypeCommand, Name: "commits", Purpose: "获取最近提交供 CI 关联分析", Target: "commit +list --limit 30"},
+			{Type: wf.StepTypeCommand, Name: "branches", Purpose: "获取分支列表检查保护状态", Target: "branch +list"},
+			{Type: wf.StepTypeSkill, Name: "review", Purpose: "AI 审查 PR 代码质量", Target: "gitlink-review", DependsOn: []string{"open-prs"}},
+			{Type: wf.StepTypeSkill, Name: "ci-diagnosis", Purpose: "AI 诊断 CI 构建失败并给出建议", Target: "gitlink-ci", DependsOn: []string{"ci-builds", "commits"}},
+		},
+	})
+}

shortcuts/workflow/defs/community_ops.go

@@ -0,0 +1,29 @@
+package defs
+
+import wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+
+// RegisterCommunityOps registers the community-ops workflow definition.
+func RegisterCommunityOps() {
+	wf.Register(&wf.WorkflowDef{
+		Name:        "community-ops",
+		Category:    "运营",
+		Description: "社区运营自动化：Issue 智能分拣 → 生成周报 → 生成 Release Notes",
+		Trigger: wf.TriggerDef{
+			Type:     "poll",
+			On:       "issue.created",
+			Interval: "5m",
+		},
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeCommand, Name: "open-issues", Purpose: "获取所有开放 Issue 供 AI 分类", Target: "issue +list --state open --limit 50"},
+			{Type: wf.StepTypeCommand, Name: "labels", Purpose: "获取标签库供 AI 匹配", Target: "label +list"},
+			{Type: wf.StepTypeCommand, Name: "members", Purpose: "获取成员列表供 AI 分配责任人", Target: "member +list"},
+			{Type: wf.StepTypeSkill, Name: "triage", Purpose: "AI 分析前三步数据，输出分拣表格并执行打标签/分配", Target: "gitlink-triage", DependsOn: []string{"open-issues", "labels", "members"}, RunWhen: wf.RunAlways},
+			{Type: wf.StepTypeCommand, Name: "repo-info", Purpose: "获取项目基础信息", Target: "repo +info"},
+			{Type: wf.StepTypeCommand, Name: "merged-prs", Purpose: "获取已合并 PR 计算合并效率", Target: "pr +list --state merged --limit 50"},
+			{Type: wf.StepTypeCommand, Name: "commits", Purpose: "获取提交历史分析活跃度", Target: "commit +list --limit 50"},
+			{Type: wf.StepTypeSkill, Name: "health-report", Purpose: "AI 根据指标生成周报", Target: "gitlink-health", DependsOn: []string{"repo-info", "merged-prs", "commits", "open-issues"}, RunWhen: wf.RunWeekly},
+			{Type: wf.StepTypeCommand, Name: "releases", Purpose: "获取版本发布记录", Target: "release +list"},
+			{Type: wf.StepTypeSkill, Name: "changelog", Purpose: "AI 分类 commit 生成 Release Notes", Target: "gitlink-changelog", DependsOn: []string{"commits", "releases", "merged-prs"}, RunWhen: wf.RunOnChange},
+		},
+	})
+}

shortcuts/workflow/defs/contributor_growth.go

@@ -0,0 +1,26 @@
+package defs
+
+import wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+
+// RegisterContributorGrowth registers the contributor-growth workflow definition.
+func RegisterContributorGrowth() {
+	wf.Register(&wf.WorkflowDef{
+		Name:        "contributor-growth",
+		Category:    "成长",
+		Description: "贡献者成长体系：追踪贡献者活动 → 生成排行 → 识别活跃与流失",
+		Trigger: wf.TriggerDef{
+			Type:     "cron",
+			On:       "0 9 * * 1",
+			Interval: "24h",
+		},
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeCommand, Name: "commits", Purpose: "提交历史统计代码贡献", Target: "commit +list --limit 100"},
+			{Type: wf.StepTypeCommand, Name: "open-issues", Purpose: "开放 Issue 统计 Issue 贡献", Target: "issue +list --state open --limit 50"},
+			{Type: wf.StepTypeCommand, Name: "closed-issues", Purpose: "已关闭 Issue 统计解决贡献", Target: "issue +list --state closed --limit 50"},
+			{Type: wf.StepTypeCommand, Name: "merged-prs", Purpose: "已合并 PR 统计代码贡献", Target: "pr +list --state merged --limit 50"},
+			{Type: wf.StepTypeCommand, Name: "members", Purpose: "项目成员列表统计参与度", Target: "member +list"},
+			{Type: wf.StepTypeCommand, Name: "repo-info", Purpose: "项目基础数据（Fork/Star/Watch）", Target: "repo +info"},
+			{Type: wf.StepTypeSkill, Name: "contributor-ranking", Purpose: "AI 分析贡献者排行并识别活跃与流失", Target: "gitlink-contributor-ranking", DependsOn: []string{"commits", "open-issues", "closed-issues", "merged-prs", "members"}},
+		},
+	})
+}

shortcuts/workflow/defs/init.go

@@ -0,0 +1,9 @@
+package defs
+
+func init() {
+	RegisterCommunityOps()
+	RegisterCodeQuality()
+	RegisterProjectInit()
+	RegisterMultiRepo()
+	RegisterContributorGrowth()
+}

shortcuts/workflow/defs/multi_repo.go

@@ -0,0 +1,32 @@
+package defs
+
+import wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+
+// RegisterMultiRepo registers the multi-repo workflow definition.
+func RegisterMultiRepo() {
+	wf.Register(&wf.WorkflowDef{
+		Name:        "multi-repo",
+		Category:    "协同",
+		Description: "多仓库协同：跨仓库 Issue/PR 状态看板、Release 协调发布",
+		Trigger: wf.TriggerDef{
+			Type:     "cron",
+			On:       "0 9 * * 1",
+			Interval: "24h",
+		},
+		Steps: []wf.StepDef{
+			{
+				Type:    wf.StepTypeCommand,
+				Name:    "multi-repo-snapshot",
+				Purpose: "采集多个仓库的 Issue/PR/Release/Milestone 状态",
+				Target:  "workflow-internal:multi-repo-snapshot",
+			},
+			{
+				Type:      wf.StepTypeSkill,
+				Name:      "multi-repo-coordination",
+				Purpose:   "生成统一 Issue 追踪、PR 看板、Release 协调报告",
+				Target:    "gitlink-multi-repo",
+				DependsOn: []string{"multi-repo-snapshot"},
+			},
+		},
+	})
+}

shortcuts/workflow/defs/project_init.go

@@ -0,0 +1,27 @@
+package defs
+
+import wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+
+// RegisterProjectInit registers the project-init workflow definition.
+func RegisterProjectInit() {
+	wf.Register(&wf.WorkflowDef{
+		Name:        "project-init",
+		Category:    "初始化",
+		Description: "项目一键初始化：创建仓库 → 脚手架文件 → 标签/里程碑/Issue → 许可证审计 → 健康报告",
+		Trigger: wf.TriggerDef{
+			Type: "manual",
+			On:   "manual",
+		},
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeSkill, Name: "init-scaffold", Purpose: "根据描述创建仓库并初始化脚手架（README/LICENSE/.gitignore/标签/里程碑/Issue）", Target: "gitlink-init-scaffold"},
+			{Type: wf.StepTypeCommand, Name: "repo-info", Purpose: "确认仓库已创建并获取基础信息", Target: "repo +info"},
+			{Type: wf.StepTypeCommand, Name: "existing-files", Purpose: "验证 README/LICENSE 文件", Target: "file +list"},
+			{Type: wf.StepTypeCommand, Name: "labels", Purpose: "验证标签库", Target: "label +list"},
+			{Type: wf.StepTypeSkill, Name: "license-check", Purpose: "检查许可证合规并扫描敏感信息泄露", Target: "gitlink-license", DependsOn: []string{"existing-files"}},
+			{Type: wf.StepTypeCommand, Name: "milestones", Purpose: "验证里程碑", Target: "milestone +list"},
+			{Type: wf.StepTypeCommand, Name: "existing-issues", Purpose: "验证初始 Issue", Target: "issue +list --state all --limit 10"},
+			{Type: wf.StepTypeCommand, Name: "branches", Purpose: "检查分支结构", Target: "branch +list"},
+			{Type: wf.StepTypeSkill, Name: "repo-audit", Purpose: "综合审计仓库健康度", Target: "gitlink-repo", DependsOn: []string{"repo-info", "existing-files", "labels", "milestones", "branches"}},
+		},
+	})
+}

shortcuts/workflow/engine/actions.go

@@ -0,0 +1,129 @@
+package engine
+
+import (
+	"bytes"
+	"crypto/md5"
+	"encoding/hex"
+	"encoding/json"
+	"fmt"
+	"os"
+	"os/exec"
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func executeActions(ctx *common.RuntimeContext, actions []wf.AIAction) (int, []string) {
+	executed := 0
+	var errors []string
+	seen := make(map[string]bool, len(actions))
+	for _, action := range actions {
+		key := actionKey(action)
+		if seen[key] {
+			fmt.Fprintf(os.Stderr, "[workflow] skipped duplicate action: %s %s %s\n", action.Type, action.Module, action.Command)
+			continue
+		}
+		seen[key] = true
+		if !isActionAllowed(action) {
+			msg := fmt.Sprintf("blocked action: %s %s +%s", action.Type, action.Module, action.Command)
+			fmt.Fprintf(os.Stderr, "[workflow] %s\n", msg)
+			errors = append(errors, msg)
+			continue
+		}
+		if action.Type == "api" {
+			path := wf.ResolvePath(action.Path, ctx.Owner, ctx.Repo)
+			_, err := ctx.CallAPI(action.Method, path, action.Body)
+			if err != nil {
+				msg := fmt.Sprintf("api %s %s: %v", action.Method, path, err)
+				fmt.Fprintf(os.Stderr, "[workflow] %s\n", msg)
+				errors = append(errors, msg)
+				continue
+			}
+			executed++
+		} else if action.Type == "cli" {
+			args := []string{action.Module, action.Command}
+			for k, v := range action.Args {
+				args = append(args, "--"+k, v)
+			}
+			if ctx.Owner != "" {
+				args = append(args, "--owner", ctx.Owner)
+			}
+			if ctx.Repo != "" {
+				args = append(args, "--repo", ctx.Repo)
+			}
+			bin := wf.ResolveCLIBinary()
+			cmd := exec.Command(bin, args...)
+			var cliStderr bytes.Buffer
+			cmd.Stderr = &cliStderr
+			err := cmd.Run()
+			if err != nil {
+				stderrStr := strings.TrimSpace(cliStderr.String())
+				if action.Module == "repo" && action.Command == "+create" && strings.Contains(stderrStr, "已被使用") {
+					fmt.Fprintf(os.Stderr, "[workflow] repo %s already exists, reusing\n", action.Args["name"])
+					executed++
+					if name := action.Args["name"]; name != "" && ctx.Repo == "" {
+						ctx.Repo = name
+					}
+					continue
+				}
+				msg := fmt.Sprintf("cli %s: %v", strings.Join(args, " "), err)
+				if cliStderr.Len() > 0 {
+					msg += " — " + stderrStr
+				}
+				fmt.Fprintf(os.Stderr, "[workflow] %s\n", msg)
+				errors = append(errors, msg)
+				continue
+			}
+			executed++
+			if action.Module == "repo" && action.Command == "+create" {
+				if name := action.Args["name"]; name != "" && ctx.Repo == "" {
+					ctx.Repo = name
+				}
+			}
+		}
+	}
+	return executed, errors
+}
+
+func actionKey(action wf.AIAction) string {
+	raw, err := json.Marshal(action)
+	if err != nil {
+		return fmt.Sprintf("%s:%s:%s:%v:%s:%s", action.Type, action.Module, action.Command, action.Args, action.Method, action.Path)
+	}
+	sum := md5.Sum(raw)
+	return hex.EncodeToString(sum[:])
+}
+
+var allowedAPIMethods = map[string]bool{
+	"GET": true, "POST": true, "PATCH": true,
+}
+
+var allowedCLIModules = map[string]bool{
+	"issue": true, "pr": true, "release": true,
+	"wiki": true, "member": true, "label": true,
+	"milestone": true, "branch": true, "comment": true,
+	"repo": true, "file": true,
+}
+
+var blockedCLICommands = map[string]bool{
+	"+delete": true, "+remove": true, "+batch-delete": true,
+	"+fork": true, "+batch-fork": true,
+}
+
+func isActionAllowed(action wf.AIAction) bool {
+	if action.Type == "api" {
+		if !allowedAPIMethods[action.Method] {
+			return false
+		}
+	}
+	if action.Type == "cli" {
+		if !allowedCLIModules[action.Module] {
+			return false
+		}
+		if blockedCLICommands[action.Command] {
+			return false
+		}
+	}
+	return true
+}

shortcuts/workflow/engine/actions_test.go

@@ -0,0 +1,31 @@
+package engine
+
+import (
+	"testing"
+
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestActionAllowed(t *testing.T) {
+	cases := []struct {
+		name    string
+		action  wf.AIAction
+		allowed bool
+	}{
+		{"api GET", wf.AIAction{Type: "api", Method: "GET"}, true},
+		{"api POST", wf.AIAction{Type: "api", Method: "POST"}, true},
+		{"api PATCH", wf.AIAction{Type: "api", Method: "PATCH"}, true},
+		{"api DELETE blocked", wf.AIAction{Type: "api", Method: "DELETE"}, false},
+		{"cli issue comment", wf.AIAction{Type: "cli", Module: "issue", Command: "+comment"}, true},
+		{"cli delete blocked", wf.AIAction{Type: "cli", Module: "repo", Command: "+delete"}, false},
+		{"cli fork blocked", wf.AIAction{Type: "cli", Module: "repo", Command: "+fork"}, false},
+		{"cli repo module blocked", wf.AIAction{Type: "cli", Module: "org", Command: "+list"}, false},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := isActionAllowed(tc.action); got != tc.allowed {
+				t.Errorf("isActionAllowed(%+v) = %v, want %v", tc.action, got, tc.allowed)
+			}
+		})
+	}
+}

shortcuts/workflow/engine/engine_test.go

@@ -0,0 +1,254 @@
+package engine
+
+import (
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/gitlink-org/gitlink-cli/internal/client"
+	"github.com/gitlink-org/gitlink-cli/internal/output"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestRunWithAPISteps(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		switch {
+		case r.Method == "GET" && r.URL.Path == "/v1/owner/repo/issues.json":
+			writeJSON(t, w, output.SuccessEnvelope([]map[string]interface{}{
+				{"id": 1, "subject": "bug"},
+				{"id": 2, "subject": "feature"},
+			}, nil))
+		case r.Method == "GET" && r.URL.Path == "/v1/owner/repo/labels.json":
+			writeJSON(t, w, output.SuccessEnvelope([]map[string]interface{}{
+				{"id": 10, "name": "bug"},
+				{"id": 11, "name": "enhancement"},
+			}, nil))
+		default:
+			t.Fatalf("unexpected request: %s %s", r.Method, r.URL.Path)
+		}
+	}))
+	defer server.Close()
+
+	ctx := newTestContext(t, server)
+	wfDef := &wf.WorkflowDef{
+		Name: "test-api",
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeAPI, Name: "fetch-issues", Purpose: "get issues", Method: "GET", Target: "{v1}/issues"},
+			{Type: wf.StepTypeAPI, Name: "fetch-labels", Purpose: "get labels", Method: "GET", Target: "{v1}/labels"},
+		},
+	}
+
+	result, err := Run(ctx, wfDef, false)
+	if err != nil {
+		t.Fatalf("Run() failed: %v", err)
+	}
+	if result.Owner != "owner" || result.Repo != "repo" {
+		t.Fatalf("expected owner/repo = owner/repo, got %s/%s", result.Owner, result.Repo)
+	}
+	if len(result.Steps) != 2 {
+		t.Fatalf("expected 2 step results, got %d", len(result.Steps))
+	}
+	for _, sr := range result.Steps {
+		if !sr.OK {
+			t.Fatalf("step %q: expected ok=true, got error=%q", sr.Step, sr.Error)
+		}
+	}
+}
+
+func TestSkillStepReceivesUpstream(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path == "/v1/owner/repo/issues.json" {
+			writeJSON(t, w, output.SuccessEnvelope(map[string]interface{}{
+				"issues": []map[string]interface{}{{"id": 1}},
+			}, nil))
+		} else if r.URL.Path == "/v1/owner/repo/labels.json" {
+			writeJSON(t, w, output.SuccessEnvelope(map[string]interface{}{
+				"labels": []map[string]interface{}{{"name": "bug"}},
+			}, nil))
+		} else {
+			t.Fatalf("unexpected request: %s %s", r.Method, r.URL.Path)
+		}
+	}))
+	defer server.Close()
+
+	ctx := newTestContext(t, server)
+	wfDef := &wf.WorkflowDef{
+		Name: "test-skill-upstream",
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeAPI, Name: "get-issues", Purpose: "issues", Method: "GET", Target: "{v1}/issues"},
+			{Type: wf.StepTypeAPI, Name: "get-labels", Purpose: "labels", Method: "GET", Target: "{v1}/labels"},
+			{Type: wf.StepTypeSkill, Name: "ai-triage", Purpose: "triage", Target: "gitlink-triage"},
+		},
+	}
+
+	result, err := Run(ctx, wfDef, true) // dry-run to test upstream without AI
+	if err != nil {
+		t.Fatalf("Run() failed: %v", err)
+	}
+
+	skillData, ok := result.Steps[2].Data.(map[string]interface{})
+	if !ok {
+		t.Fatal("skill step data is not a map")
+	}
+	if v, _ := skillData["_dry_run"]; v != true {
+		t.Fatal("skill step should have _dry_run=true")
+	}
+	upstream, ok := skillData["_upstream"].(map[string]interface{})
+	if !ok {
+		t.Fatal("skill step missing _upstream map")
+	}
+	if _, hasIssues := upstream["get-issues"]; !hasIssues {
+		t.Fatal("_upstream missing get-issues key")
+	}
+	if _, hasLabels := upstream["get-labels"]; !hasLabels {
+		t.Fatal("_upstream missing get-labels key")
+	}
+	if skillData["_skill"] != "gitlink-triage" {
+		t.Fatalf("_skill = %q, want %q", skillData["_skill"], "gitlink-triage")
+	}
+}
+
+func TestSkillStepWithDependsOn(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		writeJSON(t, w, output.SuccessEnvelope(map[string]interface{}{"ok": true}, nil))
+	}))
+	defer server.Close()
+
+	ctx := newTestContext(t, server)
+	wfDef := &wf.WorkflowDef{
+		Name: "test-depends-on",
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeAPI, Name: "open-issues", Purpose: "issues", Method: "GET", Target: "{v1}/issues"},
+			{Type: wf.StepTypeAPI, Name: "labels", Purpose: "labels", Method: "GET", Target: "{v1}/labels"},
+			{Type: wf.StepTypeAPI, Name: "members", Purpose: "members", Method: "GET", Target: "{v1}/members"},
+			{Type: wf.StepTypeSkill, Name: "triage", Purpose: "triage", Target: "gitlink-triage",
+				DependsOn: []string{"open-issues", "labels"}},
+		},
+	}
+
+	result, err := Run(ctx, wfDef, true) // dry-run to test DependsOn filter without AI
+	if err != nil {
+		t.Fatalf("Run() failed: %v", err)
+	}
+
+	skillData, ok := result.Steps[3].Data.(map[string]interface{})
+	if !ok {
+		t.Fatal("skill step data is not a map")
+	}
+	upstream, ok := skillData["_upstream"].(map[string]interface{})
+	if !ok {
+		t.Fatal("skill step missing _upstream map")
+	}
+	if _, hasIssues := upstream["open-issues"]; !hasIssues {
+		t.Fatal("_upstream missing open-issues key")
+	}
+	if _, hasLabels := upstream["labels"]; !hasLabels {
+		t.Fatal("_upstream missing labels key")
+	}
+	if _, hasMembers := upstream["members"]; hasMembers {
+		t.Fatal("_upstream should NOT contain members (not in DependsOn)")
+	}
+}
+
+func TestRunStepFailure(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(500)
+		writeJSON(t, w, map[string]interface{}{
+			"ok": false, "error": "internal server error",
+		})
+	}))
+	defer server.Close()
+
+	ctx := newTestContext(t, server)
+	wfDef := &wf.WorkflowDef{
+		Name: "test-fail",
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeAPI, Name: "bad-step", Purpose: "will fail", Method: "GET", Target: "{v1}/bad"},
+		},
+	}
+
+	result, err := Run(ctx, wfDef, false)
+	if err != nil {
+		t.Fatalf("Run() returned error: %v (steps should fail gracefully)", err)
+	}
+	if result.Steps[0].OK {
+		t.Fatal("expected step to fail, but it passed")
+	}
+}
+
+func TestRunUnknownStepType(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		t.Fatalf("no request expected")
+	}))
+	defer server.Close()
+
+	ctx := newTestContext(t, server)
+	wfDef := &wf.WorkflowDef{
+		Name: "test-unknown",
+		Steps: []wf.StepDef{
+			{Type: wf.StepType("invalid"), Name: "bad", Purpose: "unknown", Target: "x"},
+		},
+	}
+
+	result, err := Run(ctx, wfDef, false)
+	if err != nil {
+		t.Fatalf("Run() returned error: %v", err)
+	}
+	if result.Steps[0].OK {
+		t.Fatal("unknown step type should fail")
+	}
+}
+
+func TestSkillStepDryRun(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		writeJSON(t, w, output.SuccessEnvelope(map[string]interface{}{"ok": true}, nil))
+	}))
+	defer server.Close()
+
+	ctx := newTestContext(t, server)
+	wfDef := &wf.WorkflowDef{
+		Name: "test-dry-run",
+		Steps: []wf.StepDef{
+			{Type: wf.StepTypeAPI, Name: "get-data", Purpose: "data", Method: "GET", Target: "{v1}/issues"},
+			{Type: wf.StepTypeSkill, Name: "ai-step", Purpose: "AI analysis", Target: "gitlink-triage",
+				DependsOn: []string{"get-data"}},
+		},
+	}
+
+	result, err := Run(ctx, wfDef, true)
+	if err != nil {
+		t.Fatalf("Run() dry-run failed: %v", err)
+	}
+
+	skillData, ok := result.Steps[1].Data.(map[string]interface{})
+	if !ok {
+		t.Fatal("skill step data is not a map")
+	}
+	if v, _ := skillData["_dry_run"]; v != true {
+		t.Fatal("dry-run skill step should have _dry_run=true")
+	}
+}
+
+func newTestContext(t *testing.T, server *httptest.Server) *common.RuntimeContext {
+	t.Helper()
+	return &common.RuntimeContext{
+		Client: &client.Client{
+			HTTP:    server.Client(),
+			BaseURL: server.URL,
+		},
+		Owner:  "owner",
+		Repo:   "repo",
+		Format: "json",
+		Args:   map[string]string{},
+	}
+}
+
+func writeJSON(t *testing.T, w http.ResponseWriter, payload interface{}) {
+	t.Helper()
+	w.Header().Set("Content-Type", "application/json")
+	if err := json.NewEncoder(w).Encode(payload); err != nil {
+		t.Fatalf("failed to write response: %v", err)
+	}
+}

shortcuts/workflow/engine/run.go

@@ -0,0 +1,175 @@
+package engine
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/state"
+)
+
+// Run executes every step in a workflow sequentially.
+func Run(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, dryRun bool) (*wf.WorkflowResult, error) {
+	return RunWithMode(ctx, wfDef, dryRun, "")
+}
+
+// RunWithMode executes a workflow with explicit AI mode control.
+func RunWithMode(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, dryRun bool, aiMode string) (*wf.WorkflowResult, error) {
+	if aiMode != "" {
+		ctx.AIMode = aiMode
+	}
+	return run(ctx, wfDef, dryRun)
+}
+
+func run(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef, dryRun bool) (*wf.WorkflowResult, error) {
+	if !shouldSkipOwnerRepoResolve(ctx, wfDef) {
+		if err := ctx.ResolveOwnerRepo(); err != nil {
+			return nil, fmt.Errorf("resolve repo: %w", err)
+		}
+	}
+
+	if ctx.Args == nil {
+		ctx.Args = make(map[string]string)
+	}
+	if _, ok := ctx.Args["dry_run"]; !ok && dryRun {
+		ctx.Args["dry_run"] = "true"
+	}
+	ctx.Args["__wf_name"] = wfDef.Name
+
+	if ctx.Owner != "" {
+		ctx.Args["_owner"] = ctx.Owner
+	}
+	if ctx.Repo != "" {
+		ctx.Args["_repo"] = ctx.Repo
+	}
+
+	var workflowState *state.WorkflowState
+	if !dryRun {
+		workflowState, _ = state.LoadState(wfDef.Name)
+	}
+
+	results := make([]wf.StepResult, 0, len(wfDef.Steps))
+	for _, step := range wfDef.Steps {
+		var upstreamHash string
+		if !dryRun && workflowState != nil {
+			upstream := collectUpstream(ctx, step)
+			upstreamHash = state.HashData(upstream)
+		}
+
+		isDaemon := ctx.Arg("daemon-loop") == "true"
+		if isDaemon && !dryRun && workflowState != nil && step.RunWhen != "" && step.RunWhen != wf.RunAlways {
+			shouldRun, skipReason := checkPhaseCondition(step, workflowState, upstreamHash)
+			if !shouldRun {
+				fmt.Fprintf(os.Stderr, "[%s] 跳过 %s: %s\n", wfDef.Name, step.Name, skipReason)
+				results = append(results, wf.StepResult{
+					Step:       step.Name,
+					Purpose:    step.Purpose,
+					Type:       step.Type,
+					OK:         true,
+					Skipped:    true,
+					SkipReason: skipReason,
+				})
+				continue
+			}
+		}
+
+		sr := ExecuteStep(ctx, step, dryRun)
+		results = append(results, *sr)
+
+		if sr.OK && !sr.Skipped && step.Name == "init-scaffold" && ctx.Repo == "" {
+			if m, ok := sr.Data.(map[string]interface{}); ok {
+				if a, ok := m["analysis"].(map[string]interface{}); ok {
+					if r, ok := a["repo"].(string); ok && r != "" {
+						parts := strings.SplitN(r, "/", 2)
+						if len(parts) == 2 {
+							ctx.Repo = parts[1]
+						}
+					}
+				}
+			}
+		}
+
+		if !dryRun && workflowState != nil && sr.OK && !sr.Skipped {
+			if workflowState.PhaseLastRun == nil {
+				workflowState.PhaseLastRun = make(map[string]string)
+			}
+			workflowState.PhaseLastRun[step.Name] = time.Now().Format(time.RFC3339)
+			if workflowState.PhaseUpstream == nil {
+				workflowState.PhaseUpstream = make(map[string]string)
+			}
+			workflowState.PhaseUpstream[step.Name] = upstreamHash
+		}
+
+		if sr.Data != nil {
+			raw, err := json.Marshal(sr.Data)
+			if err == nil {
+				ctx.Args[step.Name] = string(raw)
+			} else {
+				ctx.Args[step.Name] = fmt.Sprint(sr.Data)
+			}
+		} else if !sr.OK && sr.Error != "" {
+			ctx.Args[step.Name] = fmt.Sprintf(`{"_error": true, "_message": %q}`, sr.Error)
+		}
+	}
+
+	if !dryRun && workflowState != nil {
+		workflowState.Save()
+	}
+
+	return &wf.WorkflowResult{
+		Workflow: wfDef.Name,
+		Owner:    ctx.Owner,
+		Repo:     ctx.Repo,
+		Steps:    results,
+	}, nil
+}
+
+func shouldSkipOwnerRepoResolve(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef) bool {
+	if wfDef != nil && wfDef.Name == "project-init" {
+		return true
+	}
+	return isExplicitMultiRepoRun(ctx, wfDef)
+}
+
+// IsExplicitMultiRepoRun reports whether this is a multi-repo run with explicit
+// --repos or --from flags (as opposed to resolving from the current directory).
+func IsExplicitMultiRepoRun(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef) bool {
+	return isExplicitMultiRepoRun(ctx, wfDef)
+}
+
+func isExplicitMultiRepoRun(ctx *common.RuntimeContext, wfDef *wf.WorkflowDef) bool {
+	if wfDef == nil || wfDef.Name != "multi-repo" {
+		return false
+	}
+	return ctx.Arg("repos") != "" || ctx.Arg("from") != ""
+}
+
+func checkPhaseCondition(step wf.StepDef, workflowState *state.WorkflowState, upstreamHash string) (bool, string) {
+	switch step.RunWhen {
+	case wf.RunWeekly:
+		last := workflowState.PhaseLastRun[step.Name]
+		if last == "" {
+			return true, ""
+		}
+		t, err := time.Parse(time.RFC3339, last)
+		if err != nil {
+			return true, ""
+		}
+		if time.Since(t) >= 7*24*time.Hour {
+			return true, ""
+		}
+		next := t.Add(7 * 24 * time.Hour)
+		return false, fmt.Sprintf("下次运行: %s", next.Format("01-02 15:04"))
+	case wf.RunOnChange:
+		if prev, ok := workflowState.PhaseUpstream[step.Name]; ok && prev == upstreamHash {
+			return false, "数据无变化"
+		}
+		return true, ""
+	default:
+		return true, ""
+	}
+}

shortcuts/workflow/engine/skill.go

@@ -0,0 +1,322 @@
+package engine
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/ai"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/state"
+)
+
+func executeSkillStep(ctx *common.RuntimeContext, step wf.StepDef, sr *wf.StepResult, dryRun bool) {
+	upstream := collectUpstream(ctx, step)
+
+	if step.Target == "gitlink-review" {
+		state.EnrichWithPRDiffs(upstream, ctx.Owner, ctx.Repo)
+		state.FilterReviewedPRs(upstream, ctx)
+	}
+
+	if dryRun {
+		sr.OK = true
+		sr.Data = map[string]interface{}{
+			"_skill":      step.Target,
+			"_dry_run":    true,
+			"_depends_on": step.DependsOn,
+			"_upstream":   upstream,
+			"_hint":       "预览模式：展示将要传给 AI/规则引擎 的上游数据，不实际执行。",
+		}
+		return
+	}
+
+	aiMode := resolveAIMode(ctx)
+	client := ai.NewAIClient()
+	var aiResp *wf.AIResponse
+	var usedAI bool
+	var aiAnalysis interface{}
+
+	switch aiMode {
+	case wf.AIModeNoAI:
+		resp, err := runRuleEngine(step, upstream)
+		if err != nil {
+			sr.OK = true
+			sr.Data = map[string]interface{}{
+				"_skill":    step.Target,
+				"_needs_ai": true,
+				"_upstream": upstream,
+				"_error":    fmt.Sprintf("rule engine failed: %v", err),
+			}
+			return
+		}
+		aiResp = resp
+
+	case wf.AIModeAI:
+		if !client.HasKey() {
+			sr.OK = false
+			sr.Error = "AI 模式需要配置 API Key（设置 DEEPSEEK_API_KEY 环境变量或 config set deepseek_api_key）"
+			return
+		}
+		resp, err := callAI(client, step, upstream)
+		if err != nil {
+			sr.OK = false
+			sr.Error = fmt.Sprintf("AI 调用失败: %v", err)
+			return
+		}
+		aiResp = resp
+		usedAI = true
+
+	default:
+		if client.HasKey() {
+			resp, err := callAI(client, step, upstream)
+			if err == nil {
+				aiResp = resp
+				usedAI = true
+			} else {
+				fmt.Fprintf(os.Stderr, "[workflow] AI 调用失败，降级到规则引擎: %v\n", err)
+			}
+		}
+		if aiResp == nil {
+			resp, err := runRuleEngine(step, upstream)
+			if err != nil {
+				sr.OK = true
+				sr.Data = map[string]interface{}{
+					"_skill":    step.Target,
+					"_needs_ai": true,
+					"_upstream": upstream,
+					"_error":    fmt.Sprintf("AI 和规则引擎均失败: %v", err),
+				}
+				return
+			}
+			aiResp = resp
+		}
+	}
+
+	if usedAI && step.Target == "gitlink-triage" {
+		if ruleResp, err := runRuleEngine(step, upstream); err == nil {
+			aiAnalysis = aiResp.Analysis
+			aiResp.Actions = ruleResp.Actions
+			aiResp.Analysis = ruleResp.Analysis
+		} else {
+			fmt.Fprintf(os.Stderr, "[workflow] triage rule action fallback failed: %v\n", err)
+		}
+	}
+
+	if usedAI && step.Target == "gitlink-init-scaffold" {
+		if suggested := extractRepoNameFromAIAnalysis(aiResp.Analysis); suggested != "" {
+			upstream["_repo"] = suggested
+		}
+		if ruleResp, err := runRuleEngine(step, upstream); err == nil {
+			aiAnalysis = aiResp.Analysis
+			aiResp.Actions = ruleResp.Actions
+		} else {
+			fmt.Fprintf(os.Stderr, "[workflow] init-scaffold rule fallback failed: %v\n", err)
+		}
+	}
+
+	if usedAI && step.Name == "health-report" {
+		if ruleResp, err := runRuleEngine(step, upstream); err == nil && len(ruleResp.Actions) > 0 {
+			content := ""
+			if s, ok := aiResp.Analysis.(string); ok && s != "" {
+				content = s
+			}
+			if content == "" {
+				if b, err := json.MarshalIndent(aiResp.Analysis, "", "  "); err == nil {
+					content = "# 项目健康度报告\n\n" + string(b) + "\n"
+				}
+			}
+			for _, a := range ruleResp.Actions {
+				if a.Module == "wiki" && a.Command == "+create" {
+					if content != "" {
+						a.Args["content"] = content
+					}
+					aiResp.Actions = append(aiResp.Actions, a)
+				}
+			}
+		}
+	}
+
+	if usedAI && step.Name == "contributor-ranking" {
+		content := ""
+		if s, ok := aiResp.Analysis.(string); ok {
+			content = s
+		} else if b, err := json.MarshalIndent(aiResp.Analysis, "", "  "); err == nil {
+			content = "# 贡献者排行榜\n\n```json\n" + string(b) + "\n```\n"
+		}
+		if content != "" {
+			pageName := "贡献者排行榜 " + time.Now().Format("2006-01-02")
+			aiResp.Actions = append(aiResp.Actions,
+				wf.AIAction{
+					Type: "cli", Module: "wiki", Command: "+create",
+					Args: map[string]string{
+						"name":    pageName,
+						"content": content,
+						"message": "自动生成贡献者排行榜",
+					},
+				},
+				wf.AIAction{
+					Type: "cli", Module: "wiki", Command: "+update",
+					Args: map[string]string{
+						"name":    pageName,
+						"content": content,
+						"message": "自动更新贡献者排行榜",
+					},
+				},
+			)
+		}
+	}
+
+	if usedAI && step.Name == "multi-repo-coordination" {
+		if ruleResp, err := runRuleEngine(step, upstream); err == nil && len(ruleResp.Actions) > 0 {
+			content := ""
+			if s, ok := aiResp.Analysis.(string); ok && s != "" {
+				content = s
+			}
+			if content == "" {
+				if b, err := json.MarshalIndent(aiResp.Analysis, "", "  "); err == nil {
+					content = "# 多仓库协同报告\n\n```json\n" + string(b) + "\n```\n"
+				}
+			}
+			for _, a := range ruleResp.Actions {
+				if a.Module == "wiki" && a.Command == "+create" {
+					if content != "" {
+						a.Args["content"] = content
+					}
+					aiResp.Actions = append(aiResp.Actions, a)
+				}
+			}
+		}
+	}
+
+	executed, actionErrors := executeActions(ctx, aiResp.Actions)
+
+	if step.Target == "gitlink-review" && !dryRun {
+		state.SaveReviewedPRFingerprints(upstream, ctx)
+	}
+
+	sr.OK = executed > 0 || len(aiResp.Actions) == 0
+	data := map[string]interface{}{
+		"ok":       sr.OK,
+		"analysis": aiResp.Analysis,
+		"executed": executed,
+		"_ai_used": usedAI,
+		"_skill":   step.Target,
+	}
+	if len(actionErrors) > 0 {
+		data["errors"] = actionErrors
+	}
+	if aiAnalysis != nil {
+		data["ai_analysis"] = aiAnalysis
+	}
+	sr.Data = data
+}
+
+func resolveAIMode(ctx *common.RuntimeContext) wf.AIMode {
+	switch ctx.AIMode {
+	case "ai":
+		return wf.AIModeAI
+	case "no-ai":
+		return wf.AIModeNoAI
+	default:
+		return wf.AIModeAuto
+	}
+}
+
+func callAI(client *ai.AIClient, step wf.StepDef, upstream map[string]interface{}) (*wf.AIResponse, error) {
+	skillMD := readSkillDoc(step.Target)
+	upstreamJSON, _ := json.MarshalIndent(upstream, "", "  ")
+	return client.Analyze(&wf.AIRequest{
+		SystemPrompt: skillMD,
+		UserData:     string(upstreamJSON),
+	})
+}
+
+func extractRepoNameFromAIAnalysis(analysis interface{}) string {
+	m, ok := analysis.(map[string]interface{})
+	if !ok {
+		return ""
+	}
+	for _, key := range []string{"repo_name", "repo", "_repo", "name", "suggested_name"} {
+		if v, ok := m[key].(string); ok && v != "" {
+			v = strings.ToLower(strings.TrimSpace(v))
+			v = regexp.MustCompile(`[^a-z0-9-]+`).ReplaceAllString(v, "-")
+			v = regexp.MustCompile(`-+`).ReplaceAllString(v, "-")
+			v = strings.Trim(v, "-")
+			if len(v) >= 2 {
+				return v
+			}
+		}
+	}
+	return ""
+}
+
+func runRuleEngine(step wf.StepDef, upstream map[string]interface{}) (*wf.AIResponse, error) {
+	engine, ok := wf.RuleEngines[step.Target]
+	if !ok {
+		return nil, wf.ErrNoRuleEngine(step.Target)
+	}
+	return engine(upstream, step.Name)
+}
+
+func collectUpstream(ctx *common.RuntimeContext, step wf.StepDef) map[string]interface{} {
+	upstream := make(map[string]interface{})
+
+	for k, v := range ctx.Args {
+		if strings.HasPrefix(k, "_") {
+			upstream[k] = v
+		}
+	}
+
+	for _, dep := range step.DependsOn {
+		if v, ok := ctx.Args[dep]; ok {
+			var parsed interface{}
+			if err := json.Unmarshal([]byte(v), &parsed); err == nil {
+				upstream[dep] = parsed
+			} else {
+				upstream[dep] = v
+			}
+		}
+	}
+	if len(step.DependsOn) == 0 {
+		for k, v := range ctx.Args {
+			var parsed interface{}
+			if err := json.Unmarshal([]byte(v), &parsed); err == nil {
+				upstream[k] = parsed
+			} else {
+				upstream[k] = v
+			}
+		}
+	}
+	return upstream
+}
+
+func readSkillDoc(target string) string {
+	paths := []string{}
+	if exe, err := os.Executable(); err == nil {
+		exeDir := filepath.Dir(exe)
+		paths = append(paths, filepath.Join(exeDir, "skills", target, "SKILL.md"))
+		paths = append(paths, filepath.Join(exeDir, "shortcuts", "workflow", "skills", target, "SKILL.md"))
+	}
+	paths = append(paths,
+		filepath.Join("skills", target, "SKILL.md"),
+		filepath.Join("shortcuts", "workflow", "skills", target, "SKILL.md"),
+		filepath.Join("/etc/gitlink-cli/skills", target, "SKILL.md"),
+	)
+	home, err := os.UserHomeDir()
+	if err == nil {
+		paths = append(paths, filepath.Join(home, ".config", "gitlink-cli", "skills", target, "SKILL.md"))
+	}
+
+	for _, p := range paths {
+		data, err := os.ReadFile(p)
+		if err == nil {
+			return string(data)
+		}
+	}
+	return fmt.Sprintf("# %s\n\nSkill documentation not found.", target)
+}

shortcuts/workflow/engine/steps.go

@@ -0,0 +1,106 @@
+package engine
+
+import (
+	"encoding/json"
+	"fmt"
+	"os/exec"
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow/snapshot"
+)
+
+// ExecuteStep dispatches a step to the right executor based on its Type.
+func ExecuteStep(ctx *common.RuntimeContext, step wf.StepDef, dryRun bool) *wf.StepResult {
+	sr := &wf.StepResult{
+		Step:    step.Name,
+		Purpose: step.Purpose,
+		Type:    step.Type,
+	}
+
+	switch step.Type {
+	case wf.StepTypeAPI:
+		executeAPIStep(ctx, step, sr)
+	case wf.StepTypeCommand:
+		executeCommandStep(ctx, step, sr)
+	case wf.StepTypeSkill:
+		executeSkillStep(ctx, step, sr, dryRun)
+	default:
+		sr.OK = false
+		sr.Error = fmt.Sprintf("unknown step type: %q", step.Type)
+	}
+	return sr
+}
+
+func executeAPIStep(ctx *common.RuntimeContext, step wf.StepDef, sr *wf.StepResult) {
+	path := wf.ResolvePath(step.Target, ctx.Owner, ctx.Repo)
+	env, err := ctx.CallAPIWithQuery(step.Method, path, step.Query)
+	if err != nil {
+		sr.OK = false
+		sr.Error = err.Error()
+	} else {
+		sr.OK = env.OK
+		sr.Data = env.Data
+	}
+}
+
+func executeCommandStep(ctx *common.RuntimeContext, step wf.StepDef, sr *wf.StepResult) {
+	if strings.HasPrefix(step.Target, "workflow-internal:") {
+		executeInternalCommandStep(ctx, step, sr)
+		return
+	}
+
+	parts := wf.ParseCommandTarget(step.Target)
+	if len(parts) == 0 {
+		sr.OK = false
+		sr.Error = fmt.Sprintf("empty command target: %q", step.Target)
+		return
+	}
+
+	bin := wf.ResolveCLIBinary()
+
+	args := append(parts, "--format", "json")
+	if ctx.Owner != "" {
+		args = append(args, "--owner", ctx.Owner)
+	}
+	if ctx.Repo != "" {
+		args = append(args, "--repo", ctx.Repo)
+	}
+
+	cmd := exec.Command(bin, args...)
+	cmd.Stderr = nil
+
+	out, err := cmd.Output()
+	if err != nil {
+		sr.OK = false
+		sr.Error = fmt.Sprintf("command failed: %v", err)
+		return
+	}
+
+	var data interface{}
+	if err := json.Unmarshal(out, &data); err != nil {
+		sr.OK = true
+		sr.Data = strings.TrimSpace(string(out))
+	} else {
+		sr.OK = true
+		sr.Data = data
+	}
+}
+
+func executeInternalCommandStep(ctx *common.RuntimeContext, step wf.StepDef, sr *wf.StepResult) {
+	switch strings.TrimPrefix(step.Target, "workflow-internal:") {
+	case "multi-repo-snapshot":
+		snap, err := snapshot.BuildMultiRepoSnapshot(ctx)
+		if err != nil {
+			sr.OK = false
+			sr.Error = err.Error()
+			return
+		}
+		sr.OK = true
+		sr.Data = snap
+	default:
+		sr.OK = false
+		sr.Error = fmt.Sprintf("unknown internal workflow command: %q", step.Target)
+	}
+}

shortcuts/workflow/registry.go

@@ -0,0 +1,29 @@
+package workflow
+
+import "sort"
+
+var registry = map[string]*WorkflowDef{}
+
+// Register adds a workflow definition to the global registry.
+func Register(wf *WorkflowDef) {
+	registry[wf.Name] = wf
+}
+
+// All returns all registered workflows sorted by name.
+func All() []*WorkflowDef {
+	names := make([]string, 0, len(registry))
+	for n := range registry {
+		names = append(names, n)
+	}
+	sort.Strings(names)
+	result := make([]*WorkflowDef, len(names))
+	for i, n := range names {
+		result[i] = registry[n]
+	}
+	return result
+}
+
+// Get returns a workflow by name, or nil.
+func Get(name string) *WorkflowDef {
+	return registry[name]
+}

shortcuts/workflow/rules/auto_merge.go

@@ -0,0 +1,149 @@
+package rules
+
+import (
+	"fmt"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// AutoMergeRule merges PRs when code quality criteria are met.
+// Criteria: no high-severity security findings + CI is healthy.
+func AutoMergeRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	prs := extractPRs(upstream, "open-prs")
+	if len(prs) == 0 {
+		return &workflow.AIResponse{
+			Analysis: map[string]interface{}{
+				"merged": 0,
+				"reason": "没有开放的 PR",
+			},
+		}, nil
+	}
+
+	// Check CI health from ci-diagnosis step result.
+	ciHealthy := true
+	if ciData, ok := upstream["ci-diagnosis"].(map[string]interface{}); ok {
+		if ciAnalysis, ok := ciData["analysis"].(map[string]interface{}); ok {
+			ciHealthy = IsCIHealthy(ciAnalysis)
+		}
+	}
+
+	// Check for high-severity security findings from review step.
+	hasHighSeverity := false
+	if reviewData, ok := upstream["review"].(map[string]interface{}); ok {
+		if reviewAnalysis, ok := reviewData["analysis"].(map[string]interface{}); ok {
+			hasHighSeverity = hasHighSeverityFindings(reviewAnalysis)
+		}
+	}
+
+	if hasHighSeverity {
+		return &workflow.AIResponse{
+			Analysis: map[string]interface{}{
+				"merged": 0,
+				"reason": "Review 发现高危安全问题，阻止自动合并",
+			},
+		}, nil
+	}
+
+	if !ciHealthy {
+		return &workflow.AIResponse{
+			Analysis: map[string]interface{}{
+				"merged": 0,
+				"reason": "CI 构建未通过，阻止自动合并",
+			},
+		}, nil
+	}
+
+	// All checks passed — merge each open PR.
+	var actions []workflow.AIAction
+	openCount := 0
+	for _, pr := range prs {
+		// Only merge open PRs.
+		status := str(pr, "pull_request_status", "pull_request_staus", "status", "state")
+		if status != "" && status != "open" {
+			continue
+		}
+		openCount++
+		prNum := interfaceToString(pr["pull_request_number"])
+		if prNum == "" {
+			prNum = interfaceToString(pr["id"])
+		}
+		if prNum == "" {
+			prNum = interfaceToString(pr["number"])
+		}
+		if prNum == "" {
+			prNum = interfaceToString(pr["pull_request_id"])
+		}
+		if prNum == "" {
+			continue
+		}
+		actions = append(actions, workflow.AIAction{
+			Type:    "cli",
+			Module:  "pr",
+			Command: "+merge",
+			Args: map[string]string{
+				"id":     prNum,
+				"method": "squash",
+			},
+		})
+	}
+
+	if len(actions) == 0 {
+		reason := "无法解析 PR 编号"
+		if openCount == 0 {
+			reason = "没有开放的 PR"
+		}
+		return &workflow.AIResponse{
+			Analysis: map[string]interface{}{
+				"merged": 0,
+				"reason": reason,
+			},
+		}, nil
+	}
+
+	return &workflow.AIResponse{
+		Analysis: map[string]interface{}{
+			"merged": len(actions),
+			"reason": fmt.Sprintf("质量达标，已合并 %d 个 PR", len(actions)),
+		},
+		Actions: actions,
+	}, nil
+}
+
+// hasHighSeverityFindings checks if the review analysis contains any
+// high-severity security findings that should block auto-merge.
+func hasHighSeverityFindings(analysis map[string]interface{}) bool {
+	findings, ok := analysis["findings"]
+	if !ok || findings == nil {
+		return false
+	}
+
+	// Handle []finding (direct from rule engine).
+	if fList, ok := findings.([]finding); ok {
+		for _, f := range fList {
+			if f.Severity == "high" && f.Lens == "security" {
+				return true
+			}
+		}
+		return false
+	}
+
+	// Handle []interface{} (after JSON round-trip).
+	if fList, ok := findings.([]interface{}); ok {
+		for _, item := range fList {
+			if fm, ok := item.(map[string]interface{}); ok {
+				sev := ""
+				lens := ""
+				if s, ok := fm["severity"].(string); ok {
+					sev = s
+				}
+				if l, ok := fm["lens"].(string); ok {
+					lens = l
+				}
+				if sev == "high" && lens == "security" {
+					return true
+				}
+			}
+		}
+	}
+	return false
+}

shortcuts/workflow/rules/changelog.go

@@ -0,0 +1,269 @@
+package rules
+
+import (
+	"fmt"
+	"regexp"
+	"sort"
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// commitGroup holds category → commits mapping for changelog generation.
+type commitGroup struct {
+	Category string
+	Emoji    string
+	Commits  []map[string]interface{}
+}
+
+var changelogRules = []struct {
+	emoji    string
+	name     string
+	patterns []*regexp.Regexp
+}{
+	{"✨", "新功能", []*regexp.Regexp{
+		regexp.MustCompile(`^(feat|feature|add|新增|支持)(\(.+\))?!?[：:]`),
+	}},
+	{"🐛", "Bug 修复", []*regexp.Regexp{
+		regexp.MustCompile(`^(fix|bugfix|hotfix|修复|解决)(\(.+\))?!?[：:]`),
+	}},
+	{"🔧", "改进优化", []*regexp.Regexp{
+		regexp.MustCompile(`^(refactor|perf|improve|enhance|style|fmt|optimize|优化|增强|完善|调整|格式化)(\(.+\))?!?[：:]`),
+	}},
+	{"📚", "文档", []*regexp.Regexp{
+		regexp.MustCompile(`^(docs|doc|文档|README|注释)(\(.+\))?!?[：:]`),
+	}},
+	{"🧪", "测试", []*regexp.Regexp{
+		regexp.MustCompile(`^(test|tests|测试)(\(.+\))?!?[：:]`),
+	}},
+	{"🏗️", "构建/CI", []*regexp.Regexp{
+		regexp.MustCompile(`^(build|ci|chore|构建|部署|Docker)(\(.+\))?!?[：:]`),
+	}},
+}
+
+var breakingRe = regexp.MustCompile(`!:`)
+var breakingBodyRe = regexp.MustCompile(`BREAKING[ -]CHANGE`)
+
+// ChangelogRule classifies commits and generates release notes.
+func ChangelogRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	commits := extractCommits(upstream)
+	releases := extractList(upstream, "releases")
+	mergedPRs := extractPRs(upstream, "merged-prs")
+
+	groups := []commitGroup{}
+	breaking := []map[string]interface{}{}
+	uncategorized := []map[string]interface{}{}
+
+	for _, c := range commits {
+		msg := str(c, "title", "message", "commit", "subject")
+		body := str(c, "body", "description")
+		if msg == "" {
+			continue
+		}
+
+		// Check breaking change.
+		if breakingRe.MatchString(msg) || breakingBodyRe.MatchString(body) {
+			breaking = append(breaking, c)
+			continue
+		}
+
+		categorized := false
+		for i, rule := range changelogRules {
+			for _, re := range rule.patterns {
+				if re.MatchString(msg) {
+					// Extend existing group or create new.
+					found := false
+					for j, g := range groups {
+						if g.Category == rule.name {
+							groups[j].Commits = append(groups[j].Commits, c)
+							found = true
+							break
+						}
+					}
+					if !found {
+						groups = append(groups, commitGroup{
+							Category: rule.name,
+							Emoji:    rule.emoji,
+							Commits:  []map[string]interface{}{c},
+						})
+					}
+					_ = i // suppress unused
+					categorized = true
+					break
+				}
+			}
+			if categorized {
+				break
+			}
+		}
+		if !categorized {
+			uncategorized = append(uncategorized, c)
+		}
+	}
+
+	// Sort groups: features first, then bug fixes, then rest.
+	sort.SliceStable(groups, func(i, j int) bool {
+		return orderOf(groups[i].Category) < orderOf(groups[j].Category)
+	})
+
+	// Add breaking changes group at top if present.
+	if len(breaking) > 0 {
+		groups = append([]commitGroup{{
+			Category: "破坏性变更",
+			Emoji:    "⚠️",
+			Commits:  breaking,
+		}}, groups...)
+	}
+
+	// Add uncategorized at end.
+	if len(uncategorized) > 0 {
+		groups = append(groups, commitGroup{
+			Category: "其他",
+			Emoji:    "🔀",
+			Commits:  uncategorized,
+		})
+	}
+
+	// Build analysis.
+	sections := []map[string]interface{}{}
+	for _, g := range groups {
+		items := []string{}
+		for _, c := range g.Commits {
+			msg := str(c, "title", "message", "commit", "subject")
+			sha := str(c, "sha", "id", "commit_id")
+			if sha != "" && len(sha) > 7 {
+				sha = sha[:7]
+			}
+			items = append(items, fmt.Sprintf("%s %s", sha, msg))
+		}
+		sections = append(sections, map[string]interface{}{
+			"category": g.Emoji + " " + g.Category,
+			"count":    len(g.Commits),
+			"items":    items,
+		})
+	}
+
+	analysis := map[string]interface{}{
+		"total_commits": len(commits),
+		"sections":      sections,
+	}
+
+	// Build release creation action if there are categorized commits.
+	var actions []workflow.AIAction
+	if len(commits) > 0 {
+		// Determine next version tag.
+		latestTag := "v0.0.0"
+		for _, rel := range releases {
+			if t := str(rel, "tag_name", "tag", "name"); t != "" {
+				if compareTags(t, latestTag) > 0 {
+					latestTag = t
+				}
+			}
+		}
+		// Also check merged PRs for version hints.
+		for _, pr := range mergedPRs {
+			labels := str(pr, "labels")
+			if strings.Contains(labels, "release") || strings.Contains(labels, "version") {
+				// PR merged with release label — bump version.
+			}
+		}
+
+		nextTag := bumpTag(latestTag)
+		if len(breaking) > 0 {
+			nextTag = bumpMajor(latestTag)
+		}
+
+		body := buildChangelogBody(sections, nextTag)
+
+		if nextTag != latestTag {
+			actions = append(actions, workflow.AIAction{
+				Type:    "cli",
+				Module:  "release",
+				Command: "+create",
+				Args: map[string]string{
+					"tag":  nextTag,
+					"name": nextTag,
+					"body": body,
+				},
+			})
+		}
+	}
+
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+func orderOf(cat string) int {
+	order := map[string]int{
+		"破坏性变更": 0,
+		"新功能":     1,
+		"Bug 修复":   2,
+		"改进优化":    3,
+		"文档":      4,
+		"测试":      5,
+		"构建/CI":   6,
+		"其他":      7,
+	}
+	if o, ok := order[cat]; ok {
+		return o
+	}
+	return 99
+}
+
+func compareTags(a, b string) int {
+	an := normalizeTag(a)
+	bn := normalizeTag(b)
+	if an > bn {
+		return 1
+	} else if an < bn {
+		return -1
+	}
+	return 0
+}
+
+func normalizeTag(t string) string {
+	t = strings.TrimPrefix(t, "v")
+	parts := strings.Split(t, ".")
+	for len(parts) < 3 {
+		parts = append(parts, "0")
+	}
+	return strings.Join(parts, ".")
+}
+
+func bumpTag(tag string) string {
+	parts := strings.Split(normalizeTag(tag), ".")
+	if len(parts) < 3 {
+		return "v0.1.0"
+	}
+	minor := atoi(parts[1])
+	return fmt.Sprintf("v%s.%d.0", parts[0], minor+1)
+}
+
+func bumpMajor(tag string) string {
+	parts := strings.Split(normalizeTag(tag), ".")
+	if len(parts) < 1 {
+		return "v1.0.0"
+	}
+	major := atoi(parts[0])
+	return fmt.Sprintf("v%d.0.0", major+1)
+}
+
+func atoi(s string) int {
+	var n int
+	fmt.Sscanf(s, "%d", &n)
+	return n
+}
+
+func buildChangelogBody(sections []map[string]interface{}, tag string) string {
+	var b strings.Builder
+	fmt.Fprintf(&b, "# %s\n\n", tag)
+	for _, sec := range sections {
+		fmt.Fprintf(&b, "## %s (%d)\n\n", sec["category"], sec["count"])
+		if items, ok := sec["items"].([]string); ok {
+			for _, item := range items {
+				fmt.Fprintf(&b, "- %s\n", item)
+			}
+		}
+		b.WriteString("\n")
+	}
+	return b.String()
+}

shortcuts/workflow/rules/changelog_test.go

@@ -0,0 +1,105 @@
+package rules
+
+import (
+	"testing"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestChangelogConventionalCommits(t *testing.T) {
+	upstream := map[string]interface{}{
+		"commits": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"sha": "abc12345", "title": "feat: add user login"},
+				map[string]interface{}{"sha": "def12345", "title": "fix: resolve null pointer"},
+				map[string]interface{}{"sha": "ghi12345", "title": "docs: update README"},
+			},
+		},
+		"releases":   map[string]interface{}{"data": []interface{}{}},
+		"merged-prs": map[string]interface{}{"data": []interface{}{}},
+	}
+
+	resp, err := ChangelogRule(upstream, "changelog")
+	if err != nil {
+		t.Fatalf("ChangelogRule failed: %v", err)
+	}
+	if resp.Analysis == nil {
+		t.Fatal("expected non-nil Analysis")
+	}
+
+	analysis := resp.Analysis.(map[string]interface{})
+	sections := analysis["sections"].([]map[string]interface{})
+	if len(sections) < 3 {
+		t.Fatalf("expected at least 3 sections (features, bugs, docs), got %d", len(sections))
+	}
+}
+
+func TestChangelogBreakingChange(t *testing.T) {
+	upstream := map[string]interface{}{
+		"commits": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"sha": "abc12345", "title": "feat!: drop support for v1"},
+			},
+		},
+		"releases":   map[string]interface{}{"data": []interface{}{}},
+		"merged-prs": map[string]interface{}{"data": []interface{}{}},
+	}
+
+	resp, err := ChangelogRule(upstream, "changelog")
+	if err != nil {
+		t.Fatalf("ChangelogRule failed: %v", err)
+	}
+
+	analysis := resp.Analysis.(map[string]interface{})
+	sections := analysis["sections"].([]map[string]interface{})
+	if len(sections) == 0 {
+		t.Fatal("expected breaking changes section")
+	}
+	first := sections[0]
+	if cat := first["category"]; cat == nil {
+		t.Fatal("first section missing category")
+	}
+}
+
+func TestChangelogChineseKeywords(t *testing.T) {
+	upstream := map[string]interface{}{
+		"commits": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"sha": "aaa11111", "title": "新增：用户管理模块"},
+				map[string]interface{}{"sha": "bbb11111", "title": "修复：登录页面报错"},
+			},
+		},
+		"releases":   map[string]interface{}{"data": []interface{}{}},
+		"merged-prs": map[string]interface{}{"data": []interface{}{}},
+	}
+
+	resp, err := ChangelogRule(upstream, "changelog")
+	if err != nil {
+		t.Fatalf("ChangelogRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	// Should have at least 2 sections.
+	sections := analysis["sections"].([]map[string]interface{})
+	if len(sections) < 2 {
+		t.Fatalf("expected at least 2 sections, got %d", len(sections))
+	}
+}
+
+func TestChangelogNoCommits(t *testing.T) {
+	resp, err := ChangelogRule(map[string]interface{}{}, "changelog")
+	if err != nil {
+		t.Fatalf("ChangelogRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if v := analysis["total_commits"]; v.(int) != 0 {
+		t.Fatalf("expected 0 total_commits, got %v", v)
+	}
+}
+
+func TestChangelogOutputFormat(t *testing.T) {
+	resp, err := ChangelogRule(map[string]interface{}{}, "changelog")
+	if err != nil {
+		t.Fatalf("ChangelogRule failed: %v", err)
+	}
+	var _ *workflow.AIResponse = resp
+}

shortcuts/workflow/rules/ci_diagnosis.go

@@ -0,0 +1,191 @@
+package rules
+
+import (
+	"fmt"
+	"regexp"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// CIDiagnosisRule matches CI build error logs against known patterns.
+func CIDiagnosisRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	builds := extractList(upstream, "ci-builds")
+	commits := extractCommits(upstream)
+
+	type diagnosis struct {
+		BuildNumber  interface{} `json:"build_number"`
+		Status       string      `json:"status"`
+		Pattern      string      `json:"matched_pattern"`
+		Diagnosis    string      `json:"diagnosis"`
+		Suggestion   string      `json:"suggestion"`
+		RelatedSHA   string      `json:"related_commit"`
+	}
+	var diagnoses []diagnosis
+	var actions []workflow.AIAction
+
+	for _, build := range builds {
+		status := str(build, "status", "state", "result")
+		if status != "failed" && status != "failure" && status != "error" && status != "3" {
+			// Check nested: some APIs use "build" wrapper.
+			if inner, ok := build["build"].(map[string]interface{}); ok {
+				build = inner
+				status = str(build, "status", "state", "result")
+				if status != "failed" && status != "failure" && status != "error" && status != "3" {
+					continue
+				}
+			} else {
+				continue
+			}
+		}
+
+		log := str(build, "log", "logs", "output", "build_log")
+		if log == "" {
+			continue
+		}
+
+		d := diagnoseLog(log)
+		buildNum := build["build_number"]
+		if buildNum == nil {
+			buildNum = build["id"]
+		}
+
+		// Find related commit.
+		relatedSHA := ""
+		for _, c := range commits {
+			cSha := str(c, "sha", "id", "commit_id")
+			if cSha != "" && containsAny(str(c, "title", "message", "commit"), d.Pattern) {
+				relatedSHA = cSha
+				break
+			}
+		}
+
+		diagnoses = append(diagnoses, diagnosis{
+			BuildNumber: buildNum,
+			Status:      status,
+			Pattern:     d.Pattern,
+			Diagnosis:   d.Diagnosis,
+			Suggestion:  d.Suggestion,
+			RelatedSHA:  relatedSHA,
+		})
+
+		// Auto-retry for transient failures.
+		if d.Transient {
+			actions = append(actions, workflow.AIAction{
+				Type:   "api",
+				Method: "POST",
+				Path:   fmt.Sprintf("{v1}/builds/%v/retry", buildNum),
+				Body:   map[string]interface{}{},
+			})
+		}
+	}
+
+	if len(diagnoses) == 0 {
+		return &workflow.AIResponse{
+			Analysis: map[string]interface{}{"diagnoses": nil, "message": "no failed builds found"},
+			Actions:  nil,
+		}, nil
+	}
+
+	analysis := map[string]interface{}{
+		"diagnoses":      diagnoses,
+		"total_failures": len(diagnoses),
+	}
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+type logPattern struct {
+	Re        *regexp.Regexp
+	Pattern   string
+	Diagnosis string
+	Suggestion string
+	Transient bool
+}
+
+var ciPatterns = []logPattern{
+	{regexp.MustCompile(`cannot find package|package .* is not in`), "cannot find package",
+		"依赖缺失",
+		"检查 go.mod/package.json 确认依赖已声明",
+		false},
+	{regexp.MustCompile(`syntax error|unexpected token|unexpected EOF`), "syntax error",
+		"语法错误",
+		"检查最近提交中的语法问题",
+		false},
+	{regexp.MustCompile(`permission denied|access denied|forbidden|401|403`), "permission denied",
+		"权限不足",
+		"检查密钥配置和访问权限",
+		false},
+	{regexp.MustCompile(`connection refused|connection reset|no route to host|dial tcp`), "connection refused",
+		"服务不可达",
+		"检查外部服务状态和网络连接",
+		true},
+	{regexp.MustCompile(`out of memory|OOM|killed|signal: killed`), "out of memory",
+		"资源不足（内存溢出）",
+		"优化内存使用或增加构建资源",
+		false},
+	{regexp.MustCompile(`No such file|file not found|not found`), "No such file",
+		"文件缺失",
+		"检查 .devops/ 路径和依赖文件配置",
+		false},
+	{regexp.MustCompile(`docker:.*not found|docker.*command not found`), "docker not found",
+		"Docker 环境缺失",
+		"构建环境未配置 Docker，检查 CI 配置",
+		false},
+	{regexp.MustCompile(`FAIL|exit status [1-9]|Test.*failed`), "exit status 1",
+		"测试失败",
+		"查看测试输出定位失败用例",
+		false},
+	{regexp.MustCompile(`timeout|timed out|deadline exceeded`), "timeout",
+		"构建超时",
+		"优化构建脚本或增加超时时间",
+		true},
+	{regexp.MustCompile(`undefined:|undefined symbol|cannot use|type mismatch`), "undefined:",
+		"编译错误（未定义符号）",
+		"检查导入和类型定义",
+		false},
+}
+
+func diagnoseLog(log string) logPattern {
+	for _, p := range ciPatterns {
+		if p.Re.MatchString(log) {
+			return p
+		}
+	}
+	return logPattern{
+		Pattern:    "unknown",
+		Diagnosis:  "未知错误",
+		Suggestion: "请人工查看 CI 日志进行诊断",
+		Transient:  false,
+	}
+}
+
+func containsAny(s string, patterns ...string) bool {
+	for _, p := range patterns {
+		if p != "" && len(s) > 0 && len(p) > 0 {
+			// Simple substring check.
+			if len(s) >= len(p) {
+				for i := 0; i <= len(s)-len(p); i++ {
+					if s[i:i+len(p)] == p {
+						return true
+					}
+				}
+			}
+		}
+	}
+	return false
+}
+
+// IsCIHealthy checks whether CI diagnosis results indicate no failures.
+func IsCIHealthy(analysis map[string]interface{}) bool {
+	if msg, ok := analysis["message"].(string); ok && msg == "no failed builds found" {
+		return true
+	}
+	if tf, ok := analysis["total_failures"]; ok {
+		switch v := tf.(type) {
+		case float64:
+			return v == 0
+		case int:
+			return v == 0
+		}
+	}
+	return true
+}

shortcuts/workflow/rules/ci_diagnosis_test.go

@@ -0,0 +1,74 @@
+package rules
+
+import (
+	"testing"
+)
+
+func TestCIDiagnosisPatterns(t *testing.T) {
+	tests := []struct {
+		log      string
+		pattern  string
+		transient bool
+	}{
+		{"cannot find package github.com/foo/bar", "cannot find package", false},
+		{"syntax error: unexpected token at line 42", "syntax error", false},
+		{"permission denied: unable to access /tmp/build", "permission denied", false},
+		{"connection refused: dial tcp 10.0.0.1:8080", "connection refused", true},
+		{"out of memory: process killed", "out of memory", false},
+		{"No such file or directory: .devops/build.yml", "No such file", false},
+		{"FAIL: TestLogin (0.23s)", "exit status 1", false},
+		{"timeout: deadline exceeded after 300s", "timeout", true},
+		{"undefined: UserService in main.go:15", "undefined:", false},
+	}
+
+	for _, tc := range tests {
+		d := diagnoseLog(tc.log)
+		if d.Pattern != tc.pattern {
+			t.Errorf("log=%q: expected pattern %q, got %q", tc.log, tc.pattern, d.Pattern)
+		}
+		if d.Transient != tc.transient {
+			t.Errorf("log=%q: expected transient=%v, got %v", tc.log, tc.transient, d.Transient)
+		}
+	}
+}
+
+func TestCIDiagnosisNoFailures(t *testing.T) {
+	upstream := map[string]interface{}{
+		"ci-builds": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": "1", "status": "success", "log": "build passed"},
+			},
+		},
+	}
+
+	resp, err := CIDiagnosisRule(upstream, "ci-diagnosis")
+	if err != nil {
+		t.Fatalf("CIDiagnosisRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if msg := analysis["message"]; msg != "no failed builds found" {
+		t.Fatalf("expected 'no failed builds found', got %v", msg)
+	}
+}
+
+func TestCIDiagnosisWithFailures(t *testing.T) {
+	upstream := map[string]interface{}{
+		"ci-builds": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": "1", "status": "failed", "log": "connection refused"},
+			},
+		},
+		"commits": map[string]interface{}{
+			"data": []interface{}{},
+		},
+	}
+
+	resp, err := CIDiagnosisRule(upstream, "ci-diagnosis")
+	if err != nil {
+		t.Fatalf("CIDiagnosisRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if v := analysis["total_failures"]; v.(int) != 1 {
+		t.Fatalf("expected 1 failure, got %v", v)
+	}
+}

shortcuts/workflow/rules/contributor.go

@@ -0,0 +1,410 @@
+package rules
+
+import (
+	"encoding/json"
+	"fmt"
+	"sort"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// contributorEntry holds per-contributor aggregate data.
+type contributorEntry struct {
+	Login         string   `json:"login"`
+	Name          string   `json:"name"`
+	Commits       int      `json:"commits"`
+	Issues        int      `json:"issues"`
+	PRs           int      `json:"prs"`
+	Total         int      `json:"total"`
+	Trend         float64  `json:"trend"`
+	LastActivity  string   `json:"last_activity"`
+	Tags          []string `json:"tags"`
+}
+
+// ContributorRankingRule produces a contributor ranking report.
+func ContributorRankingRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	commits := extractCommits(upstream)
+	issues := append(extractIssues(upstream, "open-issues"), extractIssues(upstream, "closed-issues")...)
+	prs := extractPRs(upstream, "merged-prs")
+	members := extractMembers(upstream)
+
+	// Aggregate per login.
+	stats := map[string]*contributorEntry{}
+	for _, c := range commits {
+		login := authorLogin(c)
+		if login == "" {
+			continue
+		}
+		e := ensureEntry(stats, login, members)
+		e.Commits++
+		e.Total++
+		if ts := commitTimestamp(c); ts != "" && ts > e.LastActivity {
+			e.LastActivity = ts
+		}
+	}
+
+	for _, i := range issues {
+		login := authorLogin(i)
+		if login == "" {
+			continue
+		}
+		e := ensureEntry(stats, login, members)
+		e.Issues++
+		e.Total++
+		if ts := issueTimestamp(i); ts != "" && ts > e.LastActivity {
+			e.LastActivity = ts
+		}
+	}
+
+	for _, p := range prs {
+		login := authorLogin(p)
+		if login == "" {
+			continue
+		}
+		e := ensureEntry(stats, login, members)
+		e.PRs++
+		e.Total++
+		if ts := prTimestamp(p); ts != "" && ts > e.LastActivity {
+			e.LastActivity = ts
+		}
+	}
+
+	// Calculate trends using 30-day windows.
+	now := time.Now()
+	cutoff30 := now.Add(-30 * 24 * time.Hour)
+	cutoff60 := now.Add(-60 * 24 * time.Hour)
+
+	recent := countInWindow(commits, cutoff30, now)
+	prev := countInWindow(commits, cutoff60, cutoff30)
+	for login := range stats {
+		rc := recent[login]
+		pc := prev[login]
+		if pc > 0 {
+			stats[login].Trend = float64(rc-pc) / float64(pc) * 100
+		} else if rc > 0 {
+			stats[login].Trend = 100
+		}
+		// Tagging.
+		if stats[login].Trend > 50 {
+			stats[login].Tags = append(stats[login].Tags, "new-star")
+		}
+		if stats[login].LastActivity != "" {
+			t, err := time.Parse(time.RFC3339, stats[login].LastActivity)
+			if err == nil && now.Sub(t) > 30*24*time.Hour {
+				stats[login].Tags = append(stats[login].Tags, "churn-risk")
+			}
+		}
+	}
+
+	// Sort by total desc.
+	entries := make([]contributorEntry, 0, len(stats))
+	for _, e := range stats {
+		entries = append(entries, *e)
+	}
+	sort.Slice(entries, func(i, j int) bool { return entries[i].Total > entries[j].Total })
+
+	// Build analysis.
+	rankings := make([]map[string]interface{}, len(entries))
+	for i, e := range entries {
+		rankings[i] = map[string]interface{}{
+			"rank":          i + 1,
+			"login":         e.Login,
+			"name":          e.Name,
+			"commits":       e.Commits,
+			"issues":        e.Issues,
+			"prs":           e.PRs,
+			"total":         e.Total,
+			"trend":         e.Trend,
+			"last_activity": e.LastActivity,
+			"tags":          e.Tags,
+		}
+	}
+
+	analysis := map[string]interface{}{
+		"title":    "贡献者排行榜",
+		"rankings": rankings,
+		"churn_risk": filterByTag(rankings, "churn-risk"),
+		"new_stars":  filterByTag(rankings, "new-star"),
+	}
+	// Build wiki page content.
+	wikiContent := buildContributorWiki(analysis, rankings)
+	pageName := "贡献者排行榜 " + time.Now().Format("2006-01-02")
+	actions := []workflow.AIAction{
+		{
+			Type: "cli", Module: "wiki", Command: "+create",
+			Args: map[string]string{
+				"name":    pageName,
+				"content": wikiContent,
+				"message": "自动生成贡献者排行榜",
+			},
+		},
+		{
+			Type: "cli", Module: "wiki", Command: "+update",
+			Args: map[string]string{
+				"name":    pageName,
+				"content": wikiContent,
+				"message": "自动更新贡献者排行榜",
+			},
+		},
+	}
+
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+func ensureEntry(stats map[string]*contributorEntry, login string, members map[string]string) *contributorEntry {
+	if e, ok := stats[login]; ok {
+		return e
+	}
+	e := &contributorEntry{Login: login, Name: members[login]}
+	stats[login] = e
+	return e
+}
+
+func filterByTag(rankings []map[string]interface{}, tag string) []map[string]interface{} {
+	var out []map[string]interface{}
+	for _, r := range rankings {
+		if tags, ok := r["tags"].([]string); ok {
+			for _, t := range tags {
+				if t == tag {
+					out = append(out, r)
+					break
+				}
+			}
+		}
+	}
+	return out
+}
+
+func countInWindow(commits []map[string]interface{}, start, end time.Time) map[string]int {
+	m := map[string]int{}
+	for _, c := range commits {
+		ts := commitTimestamp(c)
+		if ts == "" {
+			continue
+		}
+		t, err := time.Parse(time.RFC3339, ts)
+		if err != nil {
+			continue
+		}
+		if t.After(start) && t.Before(end) {
+			m[authorLogin(c)]++
+		}
+	}
+	return m
+}
+
+// --- helpers ---
+
+func extractCommits(upstream map[string]interface{}) []map[string]interface{} {
+	return extractList(upstream, "commits")
+}
+
+func extractIssues(upstream map[string]interface{}, key string) []map[string]interface{} {
+	return extractList(upstream, key)
+}
+
+func extractPRs(upstream map[string]interface{}, key string) []map[string]interface{} {
+	return extractList(upstream, key)
+}
+
+func extractMembers(upstream map[string]interface{}) map[string]string {
+	raw, ok := upstream["members"]
+	if !ok {
+		return nil
+	}
+	// Unwrap envelope: {"ok": true, "data": {"collaborators": [...]}}
+	if m, ok := raw.(map[string]interface{}); ok {
+		if data, ok := m["data"]; ok {
+			raw = data
+		}
+	}
+	// Unwrap inner key: {"members": [...]} or {"collaborators": [...]}
+	if m, ok := raw.(map[string]interface{}); ok {
+		for _, key := range []string{"members", "collaborators"} {
+			if list, ok := m[key]; ok {
+				raw = list
+				break
+			}
+		}
+	}
+	members := map[string]string{}
+	list, ok := raw.([]interface{})
+	if !ok {
+		return members
+	}
+	for _, item := range list {
+		m, ok := item.(map[string]interface{})
+		if !ok {
+			continue
+		}
+		login := str(m, "login", "username", "name")
+		id := fmt.Sprint(m["id"])
+		if login != "" && id != "" && id != "0" && id != "<nil>" {
+			members[login] = id
+		}
+	}
+	return members
+}
+
+func extractList(upstream map[string]interface{}, key string) []map[string]interface{} {
+	raw, ok := upstream[key]
+	if !ok {
+		return nil
+	}
+	// upstream values may be stored as an envelope: {"ok": true, "data": [...]}
+	if m, ok := raw.(map[string]interface{}); ok {
+		if data, ok := m["data"]; ok {
+			raw = data
+		}
+	}
+	// Some CLI commands return data as a JSON-encoded string; try to decode it.
+	if s, ok := raw.(string); ok {
+		var parsed interface{}
+		if err := json.Unmarshal([]byte(s), &parsed); err == nil {
+			raw = parsed
+		}
+	}
+	// GitLink API wraps lists inside a map: {"issues": [...], "milestones": [...], ...}
+	if m, ok := raw.(map[string]interface{}); ok {
+		for _, listKey := range []string{"commits", "issues", "pull_requests", "issue_tags", "tags", "releases", "members", "items", "milestones", "branches"} {
+			if v, ok := m[listKey]; ok {
+				if arr, ok := v.([]interface{}); ok {
+					raw = arr
+					break
+				}
+			}
+		}
+	}
+	list, _ := raw.([]interface{})
+	var out []map[string]interface{}
+	for _, item := range list {
+		if m, ok := item.(map[string]interface{}); ok {
+			out = append(out, m)
+		}
+	}
+	return out
+}
+
+func authorLogin(m map[string]interface{}) string {
+	// Try top-level keys first.
+	if s := str(m, "login", "username"); s != "" {
+		return s
+	}
+	// Try nested author/committer/user.
+	for _, key := range []string{"author", "committer", "user"} {
+		if a, ok := m[key].(map[string]interface{}); ok {
+			if s := str(a, "login", "username", "name"); s != "" {
+				return s
+			}
+		} else if s, ok := m[key].(string); ok && s != "" {
+			return s
+		}
+	}
+	return ""
+}
+
+func commitTimestamp(m map[string]interface{}) string {
+	// Try Unix timestamp (commit_time).
+	for _, key := range []string{"commit_time", "committed_date", "authored_date"} {
+		switch v := m[key].(type) {
+		case float64:
+			if v > 0 {
+				return time.Unix(int64(v), 0).UTC().Format(time.RFC3339)
+			}
+		case string:
+			if v != "" {
+				return v
+			}
+		}
+	}
+	// Try string timestamps.
+	for _, key := range []string{"created_at", "updated_at"} {
+		if s := str(m, key); s != "" {
+			return s
+		}
+	}
+	// Try nested author.
+	if a, ok := m["author"].(map[string]interface{}); ok {
+		return str(a, "date", "created_at")
+	}
+	if a, ok := m["committer"].(map[string]interface{}); ok {
+		return str(a, "date", "created_at")
+	}
+	return ""
+}
+
+func issueTimestamp(m map[string]interface{}) string {
+	return str(m, "created_at", "updated_at", "closed_at")
+}
+
+func prTimestamp(m map[string]interface{}) string {
+	return str(m, "created_at", "merged_at", "updated_at")
+}
+
+// str returns the first non-empty string value for the given keys.
+func str(m map[string]interface{}, keys ...string) string {
+	for _, k := range keys {
+		v, _ := m[k].(string)
+		if v != "" {
+			return v
+		}
+	}
+	return ""
+}
+
+// buildContributorWiki generates a markdown wiki page from ranking data.
+func buildContributorWiki(analysis map[string]interface{}, rankings []map[string]interface{}) string {
+	var sb strings.Builder
+	sb.WriteString("# 贡献者排行榜\n\n")
+	sb.WriteString(fmt.Sprintf("> 自动生成于 %s\n\n", time.Now().Format("2006-01-02 15:04")))
+
+	sb.WriteString("## 总览\n\n")
+	sb.WriteString("| 排名 | 贡献者 | 提交 | Issue | PR | 总计 | 趋势 | 标签 |\n")
+	sb.WriteString("|------|--------|------|-------|-----|------|------|------|\n")
+	for _, r := range rankings {
+		name := fmt.Sprint(r["name"])
+		if name == "" || name == "<nil>" {
+			name = fmt.Sprint(r["login"])
+		}
+		tags := ""
+		if t, ok := r["tags"].([]string); ok && len(t) > 0 {
+			tags = strings.Join(t, ", ")
+		}
+		sb.WriteString(fmt.Sprintf("| %v | %s | %v | %v | %v | %v | %.0f%% | %s |\n",
+			r["rank"], name, r["commits"], r["issues"], r["prs"], r["total"], r["trend"], tags))
+	}
+
+	// 新星
+	if newStars, ok := analysis["new_stars"].([]map[string]interface{}); ok && len(newStars) > 0 {
+		sb.WriteString("\n## 新星\n\n")
+		for _, s := range newStars {
+			name := fmt.Sprint(s["name"])
+			if name == "" || name == "<nil>" {
+				name = fmt.Sprint(s["login"])
+			}
+			sb.WriteString(fmt.Sprintf("- **%s** — 趋势 +%.0f%%\n", name, s["trend"]))
+		}
+	}
+
+	// 流失风险
+	if churn, ok := analysis["churn_risk"].([]map[string]interface{}); ok && len(churn) > 0 {
+		sb.WriteString("\n## 流失风险\n\n")
+		for _, c := range churn {
+			name := fmt.Sprint(c["name"])
+			if name == "" || name == "<nil>" {
+				name = fmt.Sprint(c["login"])
+			}
+			last := fmt.Sprint(c["last_activity"])
+			if t, err := time.Parse(time.RFC3339, last); err == nil {
+				last = t.Format("2006-01-02")
+			}
+			sb.WriteString(fmt.Sprintf("- **%s** — 最后活动 %s\n", name, last))
+		}
+	}
+
+	sb.WriteString("\n> 由 contributor-growth 工作流自动生成\n")
+
+	return sb.String()
+}

shortcuts/workflow/rules/contributor_test.go

@@ -0,0 +1,148 @@
+package rules
+
+import (
+	"testing"
+	"time"
+)
+
+func TestContributorRanking(t *testing.T) {
+	upstream := map[string]interface{}{
+		"commits": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"created_at": "2025-06-01T00:00:00Z", "author": "dev1"},
+				map[string]interface{}{"created_at": "2025-06-05T00:00:00Z", "author": "dev1"},
+				map[string]interface{}{"created_at": "2025-06-10T00:00:00Z", "author": "dev2"},
+			},
+		},
+		"open-issues":   map[string]interface{}{"data": []interface{}{}},
+		"closed-issues": map[string]interface{}{"data": []interface{}{}},
+		"merged-prs":    map[string]interface{}{"data": []interface{}{}},
+		"members": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"login": "dev1", "name": "Dev One"},
+				map[string]interface{}{"login": "dev2", "name": "Dev Two"},
+			},
+		},
+	}
+
+	resp, err := ContributorRankingRule(upstream, "contributor-ranking")
+	if err != nil {
+		t.Fatalf("ContributorRankingRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	rankings := analysis["rankings"].([]map[string]interface{})
+	if len(rankings) != 2 {
+		t.Fatalf("expected 2 rankings, got %d", len(rankings))
+	}
+
+	// dev1 should be ranked #1 (2 commits vs 1).
+	first := rankings[0]
+	if first["login"] != "dev1" {
+		t.Errorf("expected dev1 as #1, got %v", first["login"])
+	}
+	if first["rank"] != 1 {
+		t.Errorf("expected rank 1, got %v", first["rank"])
+	}
+	if first["commits"] != 2 {
+		t.Errorf("expected 2 commits, got %v", first["commits"])
+	}
+}
+
+func TestChurnRiskDetection(t *testing.T) {
+	// 35 days ago — should trigger churn risk.
+	oldDate := "2025-01-01T00:00:00Z"
+
+	upstream := map[string]interface{}{
+		"commits": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"created_at": oldDate, "author": "dev1"},
+			},
+		},
+		"open-issues":   map[string]interface{}{"data": []interface{}{}},
+		"closed-issues": map[string]interface{}{"data": []interface{}{}},
+		"merged-prs":    map[string]interface{}{"data": []interface{}{}},
+		"members": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"login": "dev1", "name": "Dev One"},
+			},
+		},
+	}
+
+	resp, err := ContributorRankingRule(upstream, "contributor-ranking")
+	if err != nil {
+		t.Fatalf("ContributorRankingRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	rankings := analysis["rankings"].([]map[string]interface{})
+	if len(rankings) > 0 {
+		tags := rankings[0]["tags"].([]string)
+		for _, tag := range tags {
+			if tag == "churn-risk" {
+				return // success
+			}
+		}
+		t.Errorf("expected churn-risk tag for old activity, got tags: %v", tags)
+	}
+}
+
+func TestContributorOutputFormat(t *testing.T) {
+	resp, err := ContributorRankingRule(map[string]interface{}{}, "contributor-ranking")
+	if err != nil {
+		t.Fatalf("ContributorRankingRule failed: %v", err)
+	}
+	if resp.Analysis == nil {
+		t.Fatal("expected non-nil Analysis")
+	}
+	if len(resp.Actions) < 1 {
+		t.Fatal("expected at least 1 wiki action")
+	}
+}
+
+func TestNewStarDetection(t *testing.T) {
+	// Use very recent dates so the commits appear in the last 30 days.
+	now := time.Now()
+	d1 := now.Add(-2 * 24 * time.Hour).Format(time.RFC3339)
+	d2 := now.Add(-3 * 24 * time.Hour).Format(time.RFC3339)
+	d3 := now.Add(-4 * 24 * time.Hour).Format(time.RFC3339)
+
+	upstream := map[string]interface{}{
+		"commits": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"created_at": d1, "author": "dev1"},
+				map[string]interface{}{"created_at": d2, "author": "dev1"},
+				map[string]interface{}{"created_at": d3, "author": "dev1"},
+			},
+		},
+		"open-issues":   map[string]interface{}{"data": []interface{}{}},
+		"closed-issues": map[string]interface{}{"data": []interface{}{}},
+		"merged-prs":    map[string]interface{}{"data": []interface{}{}},
+		"members": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"login": "dev1", "name": "Dev One"},
+			},
+		},
+	}
+
+	resp, err := ContributorRankingRule(upstream, "contributor-ranking")
+	if err != nil {
+		t.Fatalf("ContributorRankingRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	// Should have new-star or at least ranking.
+	rankings := analysis["rankings"].([]map[string]interface{})
+	if len(rankings) == 0 {
+		t.Fatal("expected at least 1 ranking entry")
+	}
+	tags, _ := rankings[0]["tags"].([]string)
+	t.Logf("tags for dev1: %v", tags)
+	// With only recent commits (no previous period), trend should be 100%, triggering new-star.
+	found := false
+	for _, tag := range tags {
+		if tag == "new-star" {
+			found = true
+		}
+	}
+	if !found {
+		t.Errorf("expected new-star tag, got: %v", tags)
+	}
+}

shortcuts/workflow/rules/dispatch.go

@@ -0,0 +1,31 @@
+package rules
+
+import "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+
+// HealthDispatchRule routes "gitlink-health" skill calls to the correct engine
+// based on step name and upstream data shape.
+func HealthDispatchRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	// contributor-ranking: from contributor_growth workflow.
+	if stepName == "contributor-ranking" {
+		return ContributorRankingRule(upstream, stepName)
+	}
+
+	// repo-health: from multi_repo workflow.
+	if stepName == "repo-health" {
+		return HealthReportRule(upstream, stepName)
+	}
+
+	// health-report: from community_ops workflow.
+	if stepName == "health-report" {
+		return HealthReportRule(upstream, stepName)
+	}
+
+	// Default: inspect upstream shape to decide.
+	// If upstream has "members" and "merged-prs" but no "repo-info", it's contributor ranking.
+	_, hasRepoInfo := upstream["repo-info"]
+	_, hasMembers := upstream["members"]
+	if hasMembers && !hasRepoInfo {
+		return ContributorRankingRule(upstream, stepName)
+	}
+	return HealthReportRule(upstream, stepName)
+}

shortcuts/workflow/rules/health_report.go

@@ -0,0 +1,413 @@
+package rules
+
+import (
+	"fmt"
+	"math"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// HealthReportRule computes a 4-dimension weighted health score.
+func HealthReportRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	repoInfo := extractFirst(upstream, "repo-info")
+	mergedPRs := extractPRs(upstream, "merged-prs")
+	commits := extractCommits(upstream)
+	openIssues := extractIssues(upstream, "open-issues")
+
+	// Compute raw metrics.
+	totalIssues := len(openIssues)               // approximation
+	totalPRs := len(mergedPRs)
+	now := time.Now()
+	recentCommits := countRecent(commits, now, 30)
+	releaseCount := 0
+	if repoInfo != nil {
+		if v, ok := repoInfo["release_count"].(float64); ok {
+			releaseCount = int(v)
+		}
+	}
+
+	// Dimension 1: Issue Health (30%)
+	issueScore := scoreIssueHealth(totalIssues, openIssues, now)
+
+	// Dimension 2: PR Health (30%)
+	prScore := scorePRHealth(totalPRs, mergedPRs, now)
+
+	// Dimension 3: Contributor Health (20%)
+	contributorScore := scoreContributorHealth(commits, now)
+
+	// Dimension 4: Activity (20%)
+	activityScore := scoreActivity(recentCommits, releaseCount, repoInfo)
+
+	// Composite score.
+	composite := issueScore*0.30 + prScore*0.30 + contributorScore*0.20 + activityScore*0.20
+
+	analysis := map[string]interface{}{
+		"title":    "项目健康度报告",
+		"composite": math.Round(composite*10) / 10,
+		"grade":    grade(composite),
+		"dimensions": map[string]interface{}{
+			"issue_health": map[string]interface{}{
+				"score":  math.Round(issueScore*10) / 10,
+				"weight": 0.30,
+				"grade":  grade(issueScore),
+			},
+			"pr_health": map[string]interface{}{
+				"score":  math.Round(prScore*10) / 10,
+				"weight": 0.30,
+				"grade":  grade(prScore),
+			},
+			"contributor_health": map[string]interface{}{
+				"score":  math.Round(contributorScore*10) / 10,
+				"weight": 0.20,
+				"grade":  grade(contributorScore),
+			},
+			"activity": map[string]interface{}{
+				"score":          math.Round(activityScore*10) / 10,
+				"weight":         0.20,
+				"grade":          grade(activityScore),
+				"recent_commits": recentCommits,
+				"releases":       releaseCount,
+			},
+		},
+	}
+	// Build wiki page action to publish the report.
+	body := buildHealthReportMarkdown(analysis, upstream)
+	pageName := fmt.Sprintf("健康度报告-%s", now.Format("2006-01-02"))
+	actions := []workflow.AIAction{
+		{
+			Type:    "cli",
+			Module:  "wiki",
+			Command: "+create",
+			Args: map[string]string{
+				"name":    pageName,
+				"content": body,
+				"message": "自动生成项目健康度报告",
+			},
+		},
+		{
+			Type:    "cli",
+			Module:  "wiki",
+			Command: "+update",
+			Args: map[string]string{
+				"name":    pageName,
+				"content": body,
+				"message": "自动更新项目健康度报告",
+			},
+		},
+	}
+
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+// buildHealthReportMarkdown renders the analysis and upstream data as a markdown report.
+func buildHealthReportMarkdown(analysis map[string]interface{}, upstream map[string]interface{}) string {
+	var b strings.Builder
+	now := time.Now()
+
+	b.WriteString("# 项目健康度报告\n\n")
+	fmt.Fprintf(&b, "> 报告生成时间：%s\n\n", now.Format("2006-01-02 15:04"))
+
+	composite, _ := analysis["composite"].(float64)
+	grade, _ := analysis["grade"].(string)
+	fmt.Fprintf(&b, "## 总体评分：%.1f 分（%s）\n\n", composite, grade)
+
+	dims, _ := analysis["dimensions"].(map[string]interface{})
+
+	b.WriteString("| 维度 | 得分 | 权重 | 等级 |\n")
+	b.WriteString("|------|------|------|------|\n")
+
+	dimDefs := []struct{ key, label string }{
+		{"issue_health", "Issue 健康度"},
+		{"pr_health", "PR 健康度"},
+		{"contributor_health", "贡献者活跃度"},
+		{"activity", "项目活跃度"},
+	}
+	for _, d := range dimDefs {
+		if dim, ok := dims[d.key].(map[string]interface{}); ok {
+			score, _ := dim["score"].(float64)
+			weight, _ := dim["weight"].(float64)
+			g, _ := dim["grade"].(string)
+			fmt.Fprintf(&b, "| %s | %.1f | %.0f%% | %s |\n", d.label, score, weight*100, g)
+		}
+	}
+	fmt.Fprintf(&b, "| **综合评分** | **%.1f** | — | **%s** |\n\n", composite, grade)
+	b.WriteString("等级标准：优秀(≥80) | 良好(60-79) | 需改进(<60)\n\n")
+
+	// Detail sections
+	b.WriteString("## 各项指标详情\n\n")
+
+	// Issue health
+	b.WriteString("### Issue 健康度\n\n")
+	openIssues := extractIssues(upstream, "open-issues")
+	fmt.Fprintf(&b, "- 开放 Issue 数：%d\n", len(openIssues))
+	staleCount := 0
+	for _, iss := range openIssues {
+		ts := issueTimestamp(iss)
+		if ts == "" {
+			continue
+		}
+		t, err := time.Parse(time.RFC3339, ts)
+		if err != nil {
+			continue
+		}
+		if now.Sub(t) > 30*24*time.Hour {
+			staleCount++
+		}
+	}
+	fmt.Fprintf(&b, "- 超 30 天未关闭 Issue 数：%d\n\n", staleCount)
+
+	// PR health
+	b.WriteString("### PR 健康度\n\n")
+	mergedPRs := extractPRs(upstream, "merged-prs")
+	fmt.Fprintf(&b, "- 已合并 PR 数：%d\n", len(mergedPRs))
+	var totalHours float64
+	prCount := 0
+	for _, pr := range mergedPRs {
+		created := prTimestamp(pr)
+		merged := str(pr, "merged_at")
+		if created == "" || merged == "" {
+			continue
+		}
+		ct, err1 := time.Parse(time.RFC3339, created)
+		mt, err2 := time.Parse(time.RFC3339, merged)
+		if err1 != nil || err2 != nil {
+			continue
+		}
+		totalHours += mt.Sub(ct).Hours()
+		prCount++
+	}
+	if prCount > 0 {
+		fmt.Fprintf(&b, "- 平均合并耗时：%.1f 天\n\n", totalHours/float64(prCount)/24)
+	} else {
+		b.WriteString("- 平均合并耗时：N/A\n\n")
+	}
+
+	// Contributor health
+	b.WriteString("### 贡献者活跃度\n\n")
+	commits := extractCommits(upstream)
+	authors := map[string]bool{}
+	for _, c := range commits {
+		ts := commitTimestamp(c)
+		if ts == "" {
+			continue
+		}
+		t, err := time.Parse(time.RFC3339, ts)
+		if err != nil {
+			continue
+		}
+		if now.Sub(t) <= 30*24*time.Hour {
+			authors[authorLogin(c)] = true
+		}
+	}
+	fmt.Fprintf(&b, "- 近 30 天活跃贡献者：%d 人\n\n", len(authors))
+
+	// Activity
+	b.WriteString("### 项目活跃度\n\n")
+	recentCommits := countRecent(commits, now, 30)
+	releaseCount := 0
+	if repoInfo := extractFirst(upstream, "repo-info"); repoInfo != nil {
+		if v, ok := repoInfo["release_count"].(float64); ok {
+			releaseCount = int(v)
+		}
+	}
+	fmt.Fprintf(&b, "- 近 30 天提交数：%d\n", recentCommits)
+	fmt.Fprintf(&b, "- 发行版本数：%d\n\n", releaseCount)
+
+	// Improvement suggestions
+	b.WriteString("## 改进建议\n\n")
+	suggestions := []string{}
+	if composite < 60 {
+		suggestions = append(suggestions, "- 项目整体健康度较低，建议重点关注以下改进方向")
+	}
+	if dim, ok := dims["issue_health"].(map[string]interface{}); ok {
+		if score, _ := dim["score"].(float64); score < 60 {
+			suggestions = append(suggestions, "- **Issue 管理**：及时关闭已解决的 Issue，减少超 30 天未响应的 Issue 堆积")
+		}
+	}
+	if dim, ok := dims["pr_health"].(map[string]interface{}); ok {
+		if score, _ := dim["score"].(float64); score < 60 {
+			suggestions = append(suggestions, "- **PR 审查**：加快 PR Review 速度，目标将平均合并时间控制在 3 天以内")
+		}
+	}
+	if dim, ok := dims["contributor_health"].(map[string]interface{}); ok {
+		if score, _ := dim["score"].(float64); score < 60 {
+			suggestions = append(suggestions, "- **社区建设**：吸引更多贡献者参与项目，可以标记 good-first-issue 降低新贡献者参与门槛")
+		}
+	}
+	if dim, ok := dims["activity"].(map[string]interface{}); ok {
+		if score, _ := dim["score"].(float64); score < 60 {
+			suggestions = append(suggestions, "- **项目活跃度**：保持定期提交和版本发布节奏，增加项目可见度")
+		}
+	}
+	if len(suggestions) == 0 {
+		suggestions = append(suggestions, "- 项目整体健康度良好，继续保持当前节奏")
+	}
+	for _, s := range suggestions {
+		b.WriteString(s)
+		b.WriteString("\n")
+	}
+
+	b.WriteString("\n> 由 community-ops 工作流自动生成\n")
+
+	return b.String()
+}
+
+func scoreIssueHealth(total int, openIssues []map[string]interface{}, now time.Time) float64 {
+	if total == 0 {
+		return 80 // neutral
+	}
+	// Stale issues: open for >30 days.
+	stale := 0
+	for _, iss := range openIssues {
+		ts := issueTimestamp(iss)
+		if ts == "" {
+			continue
+		}
+		t, err := time.Parse(time.RFC3339, ts)
+		if err != nil {
+			continue
+		}
+		if now.Sub(t) > 30*24*time.Hour {
+			stale++
+		}
+	}
+	ratio := float64(stale) / float64(max(total, 1))
+	score := (1 - ratio) * 100
+	return clamp(score)
+}
+
+func scorePRHealth(total int, mergedPRs []map[string]interface{}, now time.Time) float64 {
+	if total == 0 {
+		return 80 // neutral
+	}
+	// Avg merge time from creation.
+	var totalHours float64
+	count := 0
+	for _, pr := range mergedPRs {
+		created := prTimestamp(pr)
+		merged := str(pr, "merged_at")
+		if created == "" || merged == "" {
+			continue
+		}
+		ct, err1 := time.Parse(time.RFC3339, created)
+		mt, err2 := time.Parse(time.RFC3339, merged)
+		if err1 != nil || err2 != nil {
+			continue
+		}
+		totalHours += mt.Sub(ct).Hours()
+		count++
+	}
+	if count == 0 {
+		return 80
+	}
+	avgDays := totalHours / float64(count) / 24
+	// <3 days = excellent (100), 3-7 = good (80), >7 = needs improvement (50).
+	if avgDays < 3 {
+		return 100
+	} else if avgDays < 7 {
+		return 80
+	}
+	return 50
+}
+
+func scoreContributorHealth(commits []map[string]interface{}, now time.Time) float64 {
+	// Unique authors in last 30 days.
+	authors := map[string]bool{}
+	for _, c := range commits {
+		ts := commitTimestamp(c)
+		if ts == "" {
+			continue
+		}
+		t, err := time.Parse(time.RFC3339, ts)
+		if err != nil {
+			continue
+		}
+		if now.Sub(t) <= 30*24*time.Hour {
+			authors[authorLogin(c)] = true
+		}
+	}
+	n := len(authors)
+	// >3 active = 100, 1-3 = 60, 0 = 30.
+	if n > 3 {
+		return 100
+	} else if n >= 1 {
+		return 60
+	}
+	return 30
+}
+
+func scoreActivity(recentCommits int, releaseCount int, repoInfo map[string]interface{}) float64 {
+	score := 0.0
+	if recentCommits >= 10 {
+		score += 50
+	} else if recentCommits > 0 {
+		score += float64(recentCommits) / 10 * 50
+	}
+	if releaseCount >= 3 {
+		score += 50
+	} else if releaseCount > 0 {
+		score += float64(releaseCount) / 3 * 50
+	}
+	if score == 0 {
+		score = 30 // bare minimum if repo exists
+	}
+	return score
+}
+
+func countRecent(commits []map[string]interface{}, now time.Time, days int) int {
+	n := 0
+	for _, c := range commits {
+		ts := commitTimestamp(c)
+		if ts == "" {
+			continue
+		}
+		t, err := time.Parse(time.RFC3339, ts)
+		if err != nil {
+			continue
+		}
+		if now.Sub(t) <= time.Duration(days)*24*time.Hour {
+			n++
+		}
+	}
+	return n
+}
+
+// extractFirst returns the first map from upstream by key (some upstream data is wrapped).
+func extractFirst(upstream map[string]interface{}, key string) map[string]interface{} {
+	list := extractList(upstream, key)
+	if len(list) > 0 {
+		return list[0]
+	}
+	// Try direct map — key may contain the envelope {ok, data, ...}.
+	if m, ok := upstream[key].(map[string]interface{}); ok {
+		// Unwrap envelope if present.
+		if data, ok := m["data"]; ok {
+			if inner, ok := data.(map[string]interface{}); ok {
+				return inner
+			}
+		}
+		return m
+	}
+	return nil
+}
+
+func grade(score float64) string {
+	if score >= 80 {
+		return "优秀"
+	} else if score >= 60 {
+		return "良好"
+	}
+	return "需改进"
+}
+
+func clamp(v float64) float64 {
+	if v < 0 {
+		return 0
+	}
+	if v > 100 {
+		return 100
+	}
+	return v
+}

shortcuts/workflow/rules/health_report_test.go

@@ -0,0 +1,94 @@
+package rules
+
+import (
+	"testing"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestHealthReportScoring(t *testing.T) {
+	upstream := map[string]interface{}{
+		"repo-info": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"description": "test repo", "open_issues_count": 5, "release_count": 2},
+			},
+		},
+		"merged-prs": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{
+					"created_at": "2025-01-01T00:00:00Z",
+					"merged_at":  "2025-01-03T00:00:00Z",
+					"author":     "dev1",
+				},
+			},
+		},
+		"commits": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"created_at": "2025-06-01T00:00:00Z", "author": "dev1"},
+				map[string]interface{}{"created_at": "2025-06-05T00:00:00Z", "author": "dev1"},
+				map[string]interface{}{"created_at": "2025-06-10T00:00:00Z", "author": "dev2"},
+				map[string]interface{}{"created_at": "2025-06-15T00:00:00Z", "author": "dev3"},
+			},
+		},
+		"open-issues": map[string]interface{}{
+			"data": []interface{}{},
+		},
+	}
+
+	resp, err := HealthReportRule(upstream, "health-report")
+	if err != nil {
+		t.Fatalf("HealthReportRule failed: %v", err)
+	}
+	if resp.Analysis == nil {
+		t.Fatal("expected non-nil Analysis")
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if _, ok := analysis["composite"]; !ok {
+		t.Fatal("missing composite score")
+	}
+	if _, ok := analysis["grade"]; !ok {
+		t.Fatal("missing grade")
+	}
+	if dims, ok := analysis["dimensions"].(map[string]interface{}); !ok {
+		t.Fatal("missing dimensions")
+	} else {
+		for _, dim := range []string{"issue_health", "pr_health", "contributor_health", "activity"} {
+			if _, ok := dims[dim]; !ok {
+				t.Errorf("missing dimension: %s", dim)
+			}
+		}
+	}
+	if len(resp.Actions) != 2 {
+		t.Fatalf("expected 2 wiki actions (create + update), got %d", len(resp.Actions))
+	}
+	action := resp.Actions[0]
+	if action.Type != "cli" || action.Module != "wiki" || action.Command != "+create" {
+		t.Fatalf("expected wiki +create action, got %s %s %s", action.Type, action.Module, action.Command)
+	}
+	if action.Args["name"] == "" {
+		t.Fatal("expected non-empty wiki page name")
+	}
+	if action.Args["content"] == "" {
+		t.Fatal("expected non-empty wiki page content")
+	}
+}
+
+func TestHealthReportNoData(t *testing.T) {
+	resp, err := HealthReportRule(map[string]interface{}{}, "health-report")
+	if err != nil {
+		t.Fatalf("HealthReportRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	// Should still produce a score (neutral defaults).
+	if v := analysis["composite"]; v == nil {
+		t.Fatal("expected composite score even with no data")
+	}
+}
+
+func TestHealthReportOutputFormat(t *testing.T) {
+	resp, err := HealthReportRule(map[string]interface{}{}, "health-report")
+	if err != nil {
+		t.Fatalf("HealthReportRule failed: %v", err)
+	}
+	var _ *workflow.AIResponse = resp
+}

shortcuts/workflow/rules/init_scaffold.go

@@ -0,0 +1,319 @@
+package rules
+
+import (
+	"crypto/md5"
+	"encoding/base64"
+	"fmt"
+	"regexp"
+	"strings"
+	"time"
+	"unicode"
+	"unicode/utf8"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+const mitLicense = `MIT License
+
+Copyright (c) %d
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+`
+
+const gitignoreGo = `# Binaries
+*.exe
+*.exe~
+*.dll
+*.so
+*.dylib
+bin/
+dist/
+
+# Test binary
+*.test
+
+# Output of go coverage
+*.out
+
+# Go workspace
+go.work
+
+# IDE
+.idea/
+.vscode/
+*.swp
+*.swo
+
+# OS
+.DS_Store
+Thumbs.db
+
+# Env
+.env
+.env.local
+`
+
+type labelDef struct{ Name, Color string }
+
+var defaultLabels = []labelDef{
+	{"bug", "#d73a4a"},
+	{"enhancement", "#a2eeef"},
+	{"documentation", "#0075ca"},
+	{"good first issue", "#7057ff"},
+	{"question", "#d876e3"},
+	{"duplicate", "#cfd3d7"},
+	{"wontfix", "#ffffff"},
+}
+
+var defaultIssueTemplates = []struct {
+	Title string
+	Body  string
+}{
+	{
+		"项目初始化",
+		"# 项目初始化\n\n完成仓库基本配置和代码框架搭建。\n\n- [ ] README 文档\n- [ ] LICENSE 文件\n- [ ] .gitignore 配置\n- [ ] CI/CD 流水线",
+	},
+	{
+		"代码框架搭建",
+		"# 代码框架搭建\n\n搭建项目基本目录结构和核心代码框架。\n\n- [ ] 项目目录结构\n- [ ] 入口文件\n- [ ] 核心模块骨架",
+	},
+	{
+		"首个版本发布 v0.1.0",
+		"# v0.1.0 发布准备\n\n完成首个可用版本的开发和测试。\n\n- [ ] 核心功能开发\n- [ ] 单元测试\n- [ ] 发布说明",
+	},
+}
+
+// InitScaffoldRule creates a new repository and initializes it with standard
+// project scaffolding based on a user-supplied description.
+//
+// Upstream keys used:
+//
+//	_desc  — project description (generates repo name + README)
+//	_repo  — explicit repo name (overrides auto-generation)
+//	_owner — repository owner
+func InitScaffoldRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	desc := str(upstream, "_desc")
+	owner := str(upstream, "_owner")
+	repo := str(upstream, "_repo")
+
+	if owner == "" {
+		return nil, fmt.Errorf("missing _owner in upstream")
+	}
+
+	// Generate repo name from description if not explicitly provided.
+	if repo == "" && desc != "" {
+		repo = deriveRepoName(desc)
+	}
+	if repo == "" {
+		repo = "new-project"
+	}
+
+	// Generate README from description.
+	readme := fmt.Sprintf("# %s\n\n%s\n", repo, desc)
+	if desc == "" {
+		readme = fmt.Sprintf("# %s\n\nProject description.\n", repo)
+	}
+
+	projectDesc := desc
+	if projectDesc == "" {
+		projectDesc = repo
+	}
+
+	var actions []workflow.AIAction
+
+	// 1. Create the repository via CLI (handles user_id resolution internally).
+	actions = append(actions, workflow.AIAction{
+		Type: "cli", Module: "repo", Command: "+create",
+		Args: map[string]string{
+			"name":        repo,
+			"description": projectDesc,
+		},
+	})
+
+	// 2. README.md
+	actions = append(actions, workflow.AIAction{
+		Type: "api", Method: "POST",
+		Path: "{base}/create_file",
+		Body: map[string]interface{}{
+			"filepath": "README.md",
+			"content":  base64.StdEncoding.EncodeToString([]byte(readme)),
+			"message":  "docs: add README.md",
+			"branch":   "master",
+		},
+	})
+
+	// 3. LICENSE (MIT)
+	license := fmt.Sprintf(mitLicense, time.Now().Year())
+	actions = append(actions, workflow.AIAction{
+		Type: "api", Method: "POST",
+		Path: "{base}/create_file",
+		Body: map[string]interface{}{
+			"filepath": "LICENSE",
+			"content":  base64.StdEncoding.EncodeToString([]byte(license)),
+			"message":  "docs: add MIT LICENSE",
+			"branch":   "master",
+		},
+	})
+
+	// 4. .gitignore
+	actions = append(actions, workflow.AIAction{
+		Type: "api", Method: "POST",
+		Path: "{base}/create_file",
+		Body: map[string]interface{}{
+			"filepath": ".gitignore",
+			"content":  base64.StdEncoding.EncodeToString([]byte(gitignoreGo)),
+			"message":  "chore: add .gitignore",
+			"branch":   "master",
+		},
+	})
+
+	// 5. Default labels.
+	for _, l := range defaultLabels {
+		actions = append(actions, workflow.AIAction{
+			Type: "api", Method: "POST",
+			Path: "{v1}/issue_tags",
+			Body: map[string]interface{}{
+				"name":  l.Name,
+				"color": l.Color,
+			},
+		})
+	}
+
+	// 6. Initial milestone: v0.1.0, due 3 months from now.
+	due := time.Now().AddDate(0, 3, 0).Format("2006-01-02")
+	actions = append(actions, workflow.AIAction{
+		Type: "api", Method: "POST",
+		Path: "{v1}/milestones",
+		Body: map[string]interface{}{
+			"name":           "v0.1.0",
+			"description":    "首个版本发布",
+			"effective_date": due,
+		},
+	})
+
+	// 7. Initial issues.
+	for _, tpl := range defaultIssueTemplates {
+		actions = append(actions, workflow.AIAction{
+			Type: "api", Method: "POST",
+			Path: "{v1}/issues",
+			Body: map[string]interface{}{
+				"subject":     tpl.Title,
+				"description": tpl.Body,
+				"status_id":   1, // open
+				"priority_id": 2, // normal
+				"done_ratio":  0,
+			},
+		})
+	}
+
+	analysis := map[string]interface{}{
+		"repo":               fmt.Sprintf("%s/%s", owner, repo),
+		"description":        projectDesc,
+		"files_created":      3,
+		"labels_created":     len(defaultLabels),
+		"milestones_created": 1,
+		"issues_created":     len(defaultIssueTemplates),
+		"summary": fmt.Sprintf(
+			"仓库 %s/%s 创建完成：%d 个文件，%d 个标签，%d 个里程碑，%d 个 Issue",
+			owner, repo, 3, len(defaultLabels), 1, len(defaultIssueTemplates),
+		),
+	}
+
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+// deriveRepoName generates a short ASCII repo name from a description.
+// GitLink only allows ASCII letters, digits, underscores, hyphens, and dots
+// in repo identifiers — Chinese characters are rejected by the API.
+func deriveRepoName(desc string) string {
+	// 1. Extract English words first (handles mixed Chinese-English descriptions).
+	engWords := extractEnglishWords(desc)
+	if len(engWords) >= 2 {
+		return strings.ToLower(strings.Join(engWords[:min(3, len(engWords))], "-"))
+	}
+	if len(engWords) == 1 {
+		return strings.ToLower(engWords[0])
+	}
+
+	// 2. Strip non-ASCII characters, then sanitize what remains.
+	ascii := strings.Map(func(r rune) rune {
+		if r < 128 {
+			return r
+		}
+		return -1
+	}, desc)
+	ascii = strings.TrimSpace(ascii)
+	ascii = strings.ToLower(ascii)
+	ascii = regexp.MustCompile(`[^a-z0-9]+`).ReplaceAllString(ascii, "-")
+	ascii = strings.Trim(ascii, "-")
+
+	if len(ascii) >= 2 {
+		if len(ascii) > 30 {
+			ascii = ascii[:30]
+		}
+		return ascii
+	}
+
+	// 3. No usable ASCII content — use a stable hash-based name.
+	h := md5.Sum([]byte(desc))
+	return fmt.Sprintf("project-%x", h[:4])
+}
+
+func extractEnglishWords(s string) []string {
+	re := regexp.MustCompile(`[a-zA-Z][a-zA-Z0-9]*`)
+	words := re.FindAllString(s, -1)
+	// Filter out common stop words.
+	stop := map[string]bool{
+		"a": true, "an": true, "the": true, "is": true, "are": true,
+		"for": true, "of": true, "to": true, "in": true, "and": true,
+		"or": true, "it": true, "on": true, "at": true, "by": true,
+	}
+	var result []string
+	for _, w := range words {
+		if len(w) >= 2 && !stop[strings.ToLower(w)] {
+			result = append(result, w)
+		}
+	}
+	return result
+}
+
+func extractChinese(s string) string {
+	var result []rune
+	for _, r := range s {
+		if unicode.Is(unicode.Han, r) {
+			result = append(result, r)
+		}
+	}
+	if len(result) == 0 {
+		return ""
+	}
+	// Return at most 10 Chinese characters.
+	if len(result) > 10 {
+		result = result[:10]
+	}
+	return string(result)
+}
+
+func min(a, b int) int {
+	_ = utf8.RuneLen('a') // ensure unicode/utf8 import is used
+	if a < b {
+		return a
+	}
+	return b
+}

shortcuts/workflow/rules/init_scaffold_test.go

@@ -0,0 +1,76 @@
+package rules
+
+import (
+	"testing"
+)
+
+func TestInitScaffoldRuleActions(t *testing.T) {
+	upstream := map[string]interface{}{
+		"_owner": "testuser",
+		"_repo":  "test-project",
+		"_desc":  "A test project for CI/CD",
+	}
+	resp, err := InitScaffoldRule(upstream, "init-scaffold")
+	if err != nil {
+		t.Fatalf("InitScaffoldRule failed: %v", err)
+	}
+	if resp == nil {
+		t.Fatal("expected non-nil response")
+	}
+	if resp.Analysis == nil {
+		t.Fatal("expected analysis in response")
+	}
+
+	// Actions: 1 repo + 3 files + 7 labels + 1 milestone + 3 issues = 15
+	if len(resp.Actions) != 15 {
+		t.Errorf("expected 15 actions, got %d", len(resp.Actions))
+	}
+
+	// First action should be repo creation (CLI action).
+	if resp.Actions[0].Type != "cli" || resp.Actions[0].Command != "+create" {
+		t.Errorf("first action should be cli +create for repo creation, got type=%s command=%s",
+			resp.Actions[0].Type, resp.Actions[0].Command)
+	}
+
+	analysis := resp.Analysis.(map[string]interface{})
+	if files := analysis["files_created"].(int); files != 3 {
+		t.Errorf("files_created = %d, want 3", files)
+	}
+	if labels := analysis["labels_created"].(int); labels != 7 {
+		t.Errorf("labels_created = %d, want 7", labels)
+	}
+	if milestones := analysis["milestones_created"].(int); milestones != 1 {
+		t.Errorf("milestones_created = %d, want 1", milestones)
+	}
+	if issues := analysis["issues_created"].(int); issues != 3 {
+		t.Errorf("issues_created = %d, want 3", issues)
+	}
+}
+
+func TestInitScaffoldRuleWithDescription(t *testing.T) {
+	upstream := map[string]interface{}{
+		"_owner": "testuser",
+		"_desc":  "Docker 容器管理平台",
+	}
+	resp, err := InitScaffoldRule(upstream, "init-scaffold")
+	if err != nil {
+		t.Fatalf("InitScaffoldRule failed: %v", err)
+	}
+	if resp == nil {
+		t.Fatal("expected non-nil response")
+	}
+
+	// Repo name should be auto-generated from description.
+	analysis := resp.Analysis.(map[string]interface{})
+	repo := analysis["repo"].(string)
+	if repo == "" || repo == "testuser/new-project" {
+		t.Errorf("expected auto-generated repo name, got %q", repo)
+	}
+}
+
+func TestInitScaffoldRuleMissingOwner(t *testing.T) {
+	_, err := InitScaffoldRule(map[string]interface{}{}, "init-scaffold")
+	if err == nil {
+		t.Fatal("expected error when _owner is missing")
+	}
+}

shortcuts/workflow/rules/license.go

@@ -0,0 +1,217 @@
+package rules
+
+import (
+	"regexp"
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// riskEntry describes a single license/security risk finding.
+type riskEntry struct {
+	File    string `json:"file"`
+	Risk    string `json:"risk"`
+	Message string `json:"message"`
+}
+
+// LicenseCheckRule scans file lists and content for license compliance and sensitive data.
+func LicenseCheckRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	files := extractList(upstream, "existing-files")
+	if len(files) == 0 {
+		files = extractList(upstream, "files")
+	}
+
+	var findings []riskEntry
+	hasLicense := false
+	licenseType := ""
+
+	for _, f := range files {
+		name := str(f, "name", "filename", "path", "file_name")
+		if name == "" {
+			continue
+		}
+
+		// License file detection.
+		if isLicenseFile(name) {
+			hasLicense = true
+			content := str(f, "content", "body", "text")
+			if content != "" {
+				licenseType = detectLicenseType(content)
+			}
+		}
+
+		// Sensitive file name detection.
+		for _, fp := range filePatterns {
+			if fp.re.MatchString(strings.ToLower(name)) {
+				findings = append(findings, riskEntry{
+					File:    name,
+					Risk:    fp.risk,
+					Message: fp.message,
+				})
+			}
+		}
+
+		// Sensitive content detection.
+		content := str(f, "content", "body", "text")
+		if content != "" {
+			for _, cp := range contentPatterns {
+				if cp.re.MatchString(content) {
+					// Apply exclusion rules.
+					matches := cp.re.FindAllString(content, -1)
+					for _, match := range matches {
+						if isPlaceholder(match) {
+							continue
+						}
+						findings = append(findings, riskEntry{
+							File:    name,
+							Risk:    "high",
+							Message: cp.message + " → `" + truncate(match, 40) + "`",
+						})
+					}
+				}
+			}
+		}
+	}
+
+	// Compute scores.
+	licenseScore := 0.0
+	if hasLicense {
+		licenseScore = 100
+		if licenseType != "" {
+			licenseScore = 100
+		} else {
+			licenseScore = 70
+		}
+	}
+
+	sensitiveScore := 100.0
+	highCount := 0
+	for _, f := range findings {
+		if f.Risk == "high" {
+			highCount++
+		}
+	}
+	if highCount > 0 {
+		sensitiveScore = max(0, 100-float64(highCount)*20)
+	}
+
+	composite := licenseScore*0.35 + sensitiveScore*0.40 + 50*0.15 + 50*0.10
+
+	analysis := map[string]interface{}{
+		"has_license":   hasLicense,
+		"license_type":  licenseType,
+		"license_score": licenseScore,
+		"sensitive_score": sensitiveScore,
+		"composite_score": composite,
+		"grade":          grade(composite),
+		"findings":       findings,
+		"total_findings": len(findings),
+	}
+	return &workflow.AIResponse{Analysis: analysis, Actions: nil}, nil
+}
+
+// --- license file detection ---
+
+func isLicenseFile(name string) bool {
+	lower := strings.ToLower(name)
+	for _, pattern := range []string{"license", "copying", "notice", "licence"} {
+		if strings.Contains(lower, pattern) {
+			return true
+		}
+	}
+	return false
+}
+
+func detectLicenseType(content string) string {
+	for _, lp := range licensePatterns {
+		if lp.re.MatchString(content) {
+			return lp.name
+		}
+	}
+	return "Unknown"
+}
+
+var licensePatterns = []struct {
+	re   *regexp.Regexp
+	name string
+}{
+	{regexp.MustCompile(`(?i)MIT\s+License|Permission is hereby granted`), "MIT"},
+	{regexp.MustCompile(`(?i)Apache\s+License.*Version\s+2\.0|http://www\.apache\.org/licenses`), "Apache 2.0"},
+	{regexp.MustCompile(`(?i)GNU GENERAL PUBLIC LICENSE.*Version 3|GPL\s*v3`), "GPL v3"},
+	{regexp.MustCompile(`(?i)GNU GENERAL PUBLIC LICENSE.*Version 2|GPL\s*v2`), "GPL v2"},
+	{regexp.MustCompile(`(?i)BSD\s+(3-Clause|2-Clause|License)`), "BSD"},
+	{regexp.MustCompile(`(?i)Mulan\s+Permissive|木兰宽松许可证`), "Mulan PSL v2"},
+	{regexp.MustCompile(`(?i)Mozilla Public License|MPL`), "MPL"},
+	{regexp.MustCompile(`(?i)ISC\s+License`), "ISC"},
+	{regexp.MustCompile(`(?i)Creative Commons|CC-BY`), "Creative Commons"},
+	{regexp.MustCompile(`(?i)Unlicense|public\s+domain`), "Unlicense"},
+}
+
+// --- file pattern scanning ---
+
+type fileRiskPattern struct {
+	re      *regexp.Regexp
+	risk    string
+	message string
+}
+
+var filePatterns = []fileRiskPattern{
+	{regexp.MustCompile(`\.pem$|\.key$|\.p12$|\.pfx$`), "high", "私钥/证书文件，确认是否应纳入版本控制"},
+	{regexp.MustCompile(`id_rsa|id_dsa|id_ecdsa|id_ed25519`), "high", "SSH 私钥文件，不应提交到仓库"},
+	{regexp.MustCompile(`^\.env$|\.env\.`), "high", "环境变量文件，可能包含敏感凭据"},
+	{regexp.MustCompile(`credentials\.|\.secret$|secret\.yml`), "high", "凭据文件，可能包含敏感信息"},
+	{regexp.MustCompile(`serviceAccount\.json|\.service-account\.json`), "high", "服务账号密钥文件"},
+	{regexp.MustCompile(`.*token.*|.*secret.*`), "medium", "文件名包含 token/secret，检查内容"},
+	{regexp.MustCompile(`coverage\.out$`), "low", "覆盖率输出文件，建议添加到 .gitignore"},
+	{regexp.MustCompile(`\.exe$|\.bin$|\.dll$|\.so$`), "low", "二进制文件，检查是否应纳入版本控制"},
+	{regexp.MustCompile(`\.log$|\.tmp$`), "low", "日志/临时文件，建议添加到 .gitignore"},
+}
+
+// --- content pattern scanning ---
+
+type contentRiskPattern struct {
+	re      *regexp.Regexp
+	message string
+}
+
+var contentPatterns = []contentRiskPattern{
+	{regexp.MustCompile(`(?i)(token|api[_-]?key|apikey|secret|password|passwd|authorization)\s*[:=]\s*['"][^\s'"]{8,}['"]`),
+		"检测到硬编码凭据赋值"},
+	{regexp.MustCompile(`-----BEGIN (RSA |EC |OPENSSH |DSA )?PRIVATE KEY-----`),
+		"检测到私钥头部"},
+	{regexp.MustCompile(`(?i)GITLINK_TOKEN\s*[:=]\s*['"][^\s'"]+['"]`),
+		"检测到 GitLink Token"},
+	{regexp.MustCompile(`(?i)(mongodb|mysql|postgres|redis|jdbc)://[^\s'"]+@`),
+		"检测到数据库连接字符串"},
+	{regexp.MustCompile(`AKIA[0-9A-Z]{16}`),
+		"检测到 AWS Access Key"},
+	{regexp.MustCompile(`ghp_[a-zA-Z0-9]{36}`),
+		"检测到 GitHub 个人访问令牌"},
+	{regexp.MustCompile(`(?i)eyJ[a-zA-Z0-9_-]{10,}\.[a-zA-Z0-9_-]{10,}\.[a-zA-Z0-9_-]{10,}`),
+		"检测到 JWT 令牌格式"},
+	{regexp.MustCompile(`(?i)(\d{1,3}\.){3}\d{1,3}`),
+		"检测到硬编码 IP 地址"},
+	{regexp.MustCompile(`(?i)password\s*[:=]\s*['"]['"]`),
+		"检测到空密码"},
+}
+
+func isPlaceholder(s string) bool {
+	lower := strings.ToLower(s)
+	for _, p := range []string{"((variable))", "<token>", "your_token_here", "xxx", "replace_me", "<your", "placeholder"} {
+		if strings.Contains(lower, p) {
+			return true
+		}
+	}
+	// Skip if it looks like a format variable: {{.Var}} or ${VAR}.
+	if matched, _ := regexp.MatchString(`\{\{\.?\w+\}\}|\$\{\w+\}`, s); matched {
+		return true
+	}
+	return false
+}
+
+func truncate(s string, n int) string {
+	if len(s) <= n {
+		return s
+	}
+	return s[:n] + "..."
+}

shortcuts/workflow/rules/license_test.go

@@ -0,0 +1,101 @@
+package rules
+
+import (
+	"testing"
+)
+
+func TestLicenseFileDetection(t *testing.T) {
+	upstream := map[string]interface{}{
+		"existing-files": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"name": "src/main.go", "content": "package main"},
+				map[string]interface{}{"name": "LICENSE", "content": "MIT License\n\nPermission is hereby granted..."},
+				map[string]interface{}{"name": "README.md", "content": "# Project"},
+			},
+		},
+	}
+
+	resp, err := LicenseCheckRule(upstream, "license-check")
+	if err != nil {
+		t.Fatalf("LicenseCheckRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if v := analysis["has_license"]; v != true {
+		t.Fatal("expected has_license=true")
+	}
+	if v := analysis["license_type"]; v != "MIT" {
+		t.Fatalf("expected license_type=MIT, got %v", v)
+	}
+}
+
+func TestSensitiveContentDetection(t *testing.T) {
+	upstream := map[string]interface{}{
+		"existing-files": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"name": "config.go", "content": `api_key = "sk-1234567890abcdef"`},
+			},
+		},
+	}
+
+	resp, err := LicenseCheckRule(upstream, "license-check")
+	if err != nil {
+		t.Fatalf("LicenseCheckRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	findings := analysis["findings"].([]riskEntry)
+	if len(findings) == 0 {
+		t.Fatal("expected findings for hardcoded key")
+	}
+}
+
+func TestPlaceholderExclusion(t *testing.T) {
+	upstream := map[string]interface{}{
+		"existing-files": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"name": "config.go", "content": `api_key = "your_token_here"`},
+			},
+		},
+	}
+
+	resp, err := LicenseCheckRule(upstream, "license-check")
+	if err != nil {
+		t.Fatalf("LicenseCheckRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	findings := analysis["findings"].([]riskEntry)
+	// Placeholder should not generate riskEntry findings.
+	// But the api_key pattern may still match — check that it's excluded.
+	for _, f := range findings {
+		if f.File == "config.go" {
+			t.Logf("finding: %+v", f)
+			// Should NOT be a high risk for the api_key pattern.
+		}
+	}
+	// The placeholder exclusion should filter out the match from content patterns.
+	// But filename-based detections may still fire. Let's just check there are no
+	// high risk findings for the placeholder content.
+	for _, f := range findings {
+		if f.Risk == "high" && f.File == "config.go" {
+			t.Errorf("placeholder should be excluded, but got high risk finding: %s", f.Message)
+		}
+	}
+}
+
+func TestLicenseFileMissing(t *testing.T) {
+	upstream := map[string]interface{}{
+		"existing-files": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"name": "README.md"},
+			},
+		},
+	}
+
+	resp, err := LicenseCheckRule(upstream, "license-check")
+	if err != nil {
+		t.Fatalf("LicenseCheckRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if v := analysis["has_license"]; v != false {
+		t.Fatal("expected has_license=false")
+	}
+}

shortcuts/workflow/rules/multi_repo.go

@@ -0,0 +1,633 @@
+package rules
+
+import (
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func MultiRepoCoordinationRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	snapshot, err := extractMultiRepoSnapshot(upstream)
+	if err != nil {
+		return nil, err
+	}
+
+	now := time.Now()
+	release := mrString(snapshot["release"])
+	repos := extractSnapshotRepos(snapshot)
+
+	issueRows := make([]map[string]interface{}, 0, len(repos))
+	prRows := make([]map[string]interface{}, 0, len(repos))
+	releaseRows := make([]map[string]interface{}, 0, len(repos))
+	blockers := []map[string]interface{}{}
+	recommendations := []string{}
+	detailIssues := []map[string]interface{}{}
+	detailPRs := []map[string]interface{}{}
+
+	totalOpenIssues := 0
+	totalBlockerIssues := 0
+	totalStaleIssues := 0
+	totalHighPriorityIssues := 0
+	totalOpenPRs := 0
+	totalStalePRs := 0
+	totalConflictPRs := 0
+	alreadyReleasedRepos := 0
+
+	for _, repo := range repos {
+		owner := mrString(repo["owner"])
+		name := mrString(repo["repo"])
+		repoName := owner + "/" + name
+		if owner == "" {
+			repoName = name
+		}
+
+		issues := extractAnyList(repo["open_issues"])
+		prs := extractAnyList(repo["open_prs"])
+		releases := extractAnyList(repo["releases"])
+
+		blockerIssues := filterIssues(issues, isBlockerIssue)
+		highPriorityIssues := filterIssues(issues, isHighPriorityIssue)
+		staleIssues := filterStale(issues, now, 7)
+		stalePRs := filterStale(prs, now, 3)
+		conflictPRs := filterPRs(prs, isConflictPR)
+
+		totalOpenIssues += len(issues)
+		totalBlockerIssues += len(blockerIssues)
+		totalStaleIssues += len(staleIssues)
+		totalHighPriorityIssues += len(highPriorityIssues)
+		totalOpenPRs += len(prs)
+		totalStalePRs += len(stalePRs)
+		totalConflictPRs += len(conflictPRs)
+
+		issueRows = append(issueRows, map[string]interface{}{
+			"repo":          repoName,
+			"open":          len(issues),
+			"blockers":      len(blockerIssues),
+			"stale_7d":      len(staleIssues),
+			"high_priority": len(highPriorityIssues),
+		})
+		prRows = append(prRows, map[string]interface{}{
+			"repo":         repoName,
+			"open":         len(prs),
+			"stale_3d":     len(stalePRs),
+			"conflicts":    len(conflictPRs),
+			"needs_review": countNeedsReviewPRs(prs),
+		})
+
+		detailIssues = appendUniqueDetails(detailIssues, repoName, append(blockerIssues, highPriorityIssues...), issueDetail, now)
+		detailPRs = appendUniqueDetails(detailPRs, repoName, append(conflictPRs, stalePRs...), prDetail, now)
+
+		releaseExists := release != "" && hasTargetRelease(releases, release)
+		readyToRelease := release != "" && !releaseExists && len(blockerIssues) == 0 && len(conflictPRs) == 0 && len(stalePRs) == 0
+		releaseRows = append(releaseRows, map[string]interface{}{
+			"repo":                 repoName,
+			"target":               release,
+			"release_exists":       releaseExists,
+			"already_released":     releaseExists,
+			"ready_to_release":     readyToRelease,
+			"blocker_issues":       len(blockerIssues),
+			"open_prs":             len(prs),
+			"stale_or_conflict_pr": len(stalePRs) + len(conflictPRs),
+		})
+
+		if len(blockerIssues) > 0 {
+			blockers = append(blockers, map[string]interface{}{
+				"repo":   repoName,
+				"type":   "blocker_issues",
+				"count":  len(blockerIssues),
+				"reason": fmt.Sprintf("%s 还有 %d 个阻塞 Issue", repoName, len(blockerIssues)),
+			})
+			recommendations = append(recommendations, fmt.Sprintf("优先处理 %s 的阻塞 Issue", repoName))
+		}
+		if len(conflictPRs) > 0 {
+			blockers = append(blockers, map[string]interface{}{
+				"repo":   repoName,
+				"type":   "conflict_prs",
+				"count":  len(conflictPRs),
+				"reason": fmt.Sprintf("%s 还有 %d 个疑似冲突 PR", repoName, len(conflictPRs)),
+			})
+			recommendations = append(recommendations, fmt.Sprintf("先解决 %s 的冲突 PR", repoName))
+		}
+		if len(stalePRs) > 0 {
+			blockers = append(blockers, map[string]interface{}{
+				"repo":   repoName,
+				"type":   "stale_prs",
+				"count":  len(stalePRs),
+				"reason": fmt.Sprintf("%s 还有 %d 个超过 3 天未合并 PR", repoName, len(stalePRs)),
+			})
+		}
+		if releaseExists {
+			alreadyReleasedRepos++
+			recommendations = append(recommendations, fmt.Sprintf("%s 已存在 %s Release，确认是否属于重复发布检查", repoName, release))
+		}
+	}
+
+	errors := extractAnyList(snapshot["errors"])
+	readyToRelease := release != "" && len(blockers) == 0 && len(errors) == 0 && alreadyReleasedRepos == 0
+	if release == "" {
+		recommendations = append(recommendations, "指定 --release 可启用跨仓库发布协调检查")
+	}
+	if len(errors) > 0 {
+		blockers = append(blockers, map[string]interface{}{
+			"type":   "collection_errors",
+			"count":  len(errors),
+			"reason": fmt.Sprintf("采集过程中有 %d 个错误，需要先确认数据完整性", len(errors)),
+		})
+		recommendations = append(recommendations, "先处理采集失败的仓库或接口权限问题，再判断发布状态")
+	}
+
+	analysis := map[string]interface{}{
+		"title": "多仓库协同报告",
+		"summary": map[string]interface{}{
+			"repos":             len(repos),
+			"target_release":    release,
+			"open_issues":       totalOpenIssues,
+			"blocker_issues":    totalBlockerIssues,
+			"stale_issues_7d":   totalStaleIssues,
+			"high_priority":     totalHighPriorityIssues,
+			"open_prs":          totalOpenPRs,
+			"stale_prs_3d":      totalStalePRs,
+			"conflict_prs":      totalConflictPRs,
+			"collection_errors": len(errors),
+		},
+		"issue_tracking": map[string]interface{}{
+			"total_open":          totalOpenIssues,
+			"total_blockers":      totalBlockerIssues,
+			"total_stale_7d":      totalStaleIssues,
+			"total_high_priority": totalHighPriorityIssues,
+			"by_repo":             issueRows,
+			"details":             detailIssues,
+		},
+		"pr_board": map[string]interface{}{
+			"total_open":      totalOpenPRs,
+			"total_stale_3d":  totalStalePRs,
+			"total_conflicts": totalConflictPRs,
+			"by_repo":         prRows,
+			"details":         detailPRs,
+		},
+		"release_coordination": map[string]interface{}{
+			"target":           release,
+			"ready":            readyToRelease,
+			"ready_to_release": readyToRelease,
+			"by_repo":          releaseRows,
+			"blockers":         blockers,
+		},
+		"recommendations": uniqueStrings(recommendations),
+	}
+
+	var actions []workflow.AIAction
+	wikiOwner := mrString(upstream["_wiki_owner"])
+	wikiRepo := mrString(upstream["_wiki_repo"])
+	if wikiOwner != "" && wikiRepo != "" {
+		body := buildMultiRepoMarkdown(analysis)
+		pageName := fmt.Sprintf("多仓库协同报告-%s", now.Format("2006-01-02"))
+		actions = []workflow.AIAction{
+			{
+				Type: "cli", Module: "wiki", Command: "+create",
+				Args: map[string]string{
+					"owner":   wikiOwner,
+					"repo":    wikiRepo,
+					"name":    pageName,
+					"content": body,
+					"message": "自动生成多仓库协同报告",
+				},
+			},
+			{
+				Type: "cli", Module: "wiki", Command: "+update",
+				Args: map[string]string{
+					"owner":   wikiOwner,
+					"repo":    wikiRepo,
+					"name":    pageName,
+					"content": body,
+					"message": "自动更新多仓库协同报告",
+				},
+			},
+		}
+	}
+
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+func extractMultiRepoSnapshot(upstream map[string]interface{}) (map[string]interface{}, error) {
+	raw, ok := upstream["multi-repo-snapshot"]
+	if !ok {
+		return nil, fmt.Errorf("missing multi-repo-snapshot upstream data")
+	}
+	if m, ok := raw.(map[string]interface{}); ok {
+		if data, ok := m["data"].(map[string]interface{}); ok {
+			return data, nil
+		}
+		return m, nil
+	}
+	return nil, fmt.Errorf("invalid multi-repo-snapshot upstream data")
+}
+
+func extractSnapshotRepos(snapshot map[string]interface{}) []map[string]interface{} {
+	return extractAnyList(snapshot["repos"])
+}
+
+func extractAnyList(raw interface{}) []map[string]interface{} {
+	for i := 0; i < 3; i++ {
+		m, ok := raw.(map[string]interface{})
+		if !ok {
+			break
+		}
+		unwrapped := false
+		if data, ok := m["data"]; ok {
+			raw = data
+			unwrapped = true
+		} else {
+			for _, key := range []string{"repos", "issues", "pull_requests", "releases", "milestones", "items", "errors"} {
+				if v, ok := m[key]; ok {
+					raw = v
+					unwrapped = true
+					break
+				}
+			}
+		}
+		if !unwrapped {
+			break
+		}
+	}
+	list, _ := raw.([]interface{})
+	out := make([]map[string]interface{}, 0, len(list))
+	for _, item := range list {
+		if m, ok := item.(map[string]interface{}); ok {
+			out = append(out, m)
+		}
+	}
+	return out
+}
+
+func filterIssues(items []map[string]interface{}, pred func(map[string]interface{}) bool) []map[string]interface{} {
+	return filterMaps(items, pred)
+}
+
+func filterPRs(items []map[string]interface{}, pred func(map[string]interface{}) bool) []map[string]interface{} {
+	return filterMaps(items, pred)
+}
+
+func filterMaps(items []map[string]interface{}, pred func(map[string]interface{}) bool) []map[string]interface{} {
+	out := []map[string]interface{}{}
+	for _, item := range items {
+		if pred(item) {
+			out = append(out, item)
+		}
+	}
+	return out
+}
+
+func filterStale(items []map[string]interface{}, now time.Time, days int) []map[string]interface{} {
+	out := []map[string]interface{}{}
+	for _, item := range items {
+		t, ok := itemUpdatedAt(item)
+		if ok && now.Sub(t) >= time.Duration(days)*24*time.Hour {
+			out = append(out, item)
+		}
+	}
+	return out
+}
+
+func isBlockerIssue(issue map[string]interface{}) bool {
+	text := strings.ToLower(mrString(issue["subject"]) + " " + mrString(issue["title"]) + " " + mrString(issue["description"]) + " " + labelsText(issue))
+	return strings.Contains(text, "blocker") || strings.Contains(text, "阻塞") || strings.Contains(text, "critical") || strings.Contains(text, "严重")
+}
+
+func isHighPriorityIssue(issue map[string]interface{}) bool {
+	text := strings.ToLower(mrString(issue["subject"]) + " " + mrString(issue["title"]) + " " + mrString(issue["priority"]) + " " + labelsText(issue))
+	if strings.Contains(text, "high") || strings.Contains(text, "urgent") || strings.Contains(text, "高优先级") || strings.Contains(text, "紧急") {
+		return true
+	}
+	if id, ok := numberValue(issue["priority_id"]); ok && id >= 4 {
+		return true
+	}
+	return false
+}
+
+func isConflictPR(pr map[string]interface{}) bool {
+	text := strings.ToLower(mrString(pr["title"]) + " " + mrString(pr["body"]) + " " + mrString(pr["status"]) + " " + mrString(pr["merge_status"]))
+	return strings.Contains(text, "conflict") || strings.Contains(text, "冲突") || strings.Contains(text, "cannot merge")
+}
+
+func countNeedsReviewPRs(prs []map[string]interface{}) int {
+	count := 0
+	for _, pr := range prs {
+		text := strings.ToLower(mrString(pr["status"]) + " " + mrString(pr["review_status"]) + " " + labelsText(pr))
+		if text == "" || strings.Contains(text, "review") || strings.Contains(text, "待审") {
+			count++
+		}
+	}
+	return count
+}
+
+func hasTargetRelease(releases []map[string]interface{}, target string) bool {
+	for _, rel := range releases {
+		for _, key := range []string{"tag_name", "tag", "name", "title", "version"} {
+			if mrString(rel[key]) == target {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+func issueDetail(repo string, issue map[string]interface{}, now time.Time) map[string]interface{} {
+	days := daysSince(issue, now)
+	return map[string]interface{}{
+		"repo":       repo,
+		"number":     firstNonEmpty(issue, "number", "id", "issue_id"),
+		"title":      firstNonEmpty(issue, "subject", "title"),
+		"assignee":   assigneeName(issue),
+		"stale_days": days,
+		"blocker":    isBlockerIssue(issue),
+		"high":       isHighPriorityIssue(issue),
+	}
+}
+
+func prDetail(repo string, pr map[string]interface{}, now time.Time) map[string]interface{} {
+	return map[string]interface{}{
+		"repo":       repo,
+		"number":     firstNonEmpty(pr, "pull_request_number", "number", "id", "pull_request_id"),
+		"title":      firstNonEmpty(pr, "title", "subject"),
+		"assignee":   assigneeName(pr),
+		"stale_days": daysSince(pr, now),
+		"conflict":   isConflictPR(pr),
+	}
+}
+
+func appendUniqueDetails(
+	dst []map[string]interface{},
+	repo string,
+	items []map[string]interface{},
+	detailFn func(string, map[string]interface{}, time.Time) map[string]interface{},
+	now time.Time,
+) []map[string]interface{} {
+	seen := make(map[string]bool, len(dst)+len(items))
+	for _, item := range dst {
+		seen[detailKey(item)] = true
+	}
+	for _, item := range items {
+		detail := detailFn(repo, item, now)
+		key := detailKey(detail)
+		if seen[key] {
+			continue
+		}
+		seen[key] = true
+		dst = append(dst, detail)
+	}
+	return dst
+}
+
+func detailKey(item map[string]interface{}) string {
+	repo := mrString(item["repo"])
+	number := mrString(item["number"])
+	if number == "" {
+		number = mrString(item["title"])
+	}
+	return repo + "#" + number
+}
+
+func itemUpdatedAt(item map[string]interface{}) (time.Time, bool) {
+	for _, key := range []string{"updated_at", "updated_on", "created_at", "created_on"} {
+		switch v := item[key].(type) {
+		case string:
+			if t, ok := parseTime(v); ok {
+				return t, true
+			}
+		case float64:
+			if v > 0 {
+				return time.Unix(int64(v), 0), true
+			}
+		}
+	}
+	return time.Time{}, false
+}
+
+func parseTime(raw string) (time.Time, bool) {
+	for _, layout := range []string{time.RFC3339, "2006-01-02T15:04:05Z07:00", "2006-01-02 15:04:05", "2006-01-02"} {
+		if t, err := time.Parse(layout, raw); err == nil {
+			return t, true
+		}
+	}
+	return time.Time{}, false
+}
+
+func daysSince(item map[string]interface{}, now time.Time) int {
+	t, ok := itemUpdatedAt(item)
+	if !ok {
+		return 0
+	}
+	return int(now.Sub(t).Hours() / 24)
+}
+
+func firstNonEmpty(m map[string]interface{}, keys ...string) string {
+	for _, key := range keys {
+		if s := mrString(m[key]); s != "" {
+			return s
+		}
+	}
+	return ""
+}
+
+func labelsText(m map[string]interface{}) string {
+	var parts []string
+	for _, key := range []string{"labels", "tags", "issue_tags"} {
+		for _, label := range extractAnyList(m[key]) {
+			parts = append(parts, firstNonEmpty(label, "name", "title"))
+		}
+	}
+	return strings.Join(parts, " ")
+}
+
+func assigneeName(m map[string]interface{}) string {
+	for _, key := range []string{"assignee", "assigned_to", "user"} {
+		if nested, ok := m[key].(map[string]interface{}); ok {
+			if name := firstNonEmpty(nested, "name", "login", "username"); name != "" {
+				return name
+			}
+		}
+	}
+	return firstNonEmpty(m, "assignee", "assigned_to", "author_name")
+}
+
+func mrString(v interface{}) string {
+	if v == nil {
+		return ""
+	}
+	s := strings.TrimSpace(fmt.Sprint(v))
+	if s == "<nil>" {
+		return ""
+	}
+	return s
+}
+
+func numberValue(v interface{}) (float64, bool) {
+	switch n := v.(type) {
+	case float64:
+		return n, true
+	case int:
+		return float64(n), true
+	default:
+		return 0, false
+	}
+}
+
+func uniqueStrings(items []string) []string {
+	seen := map[string]bool{}
+	out := []string{}
+	for _, item := range items {
+		item = strings.TrimSpace(item)
+		if item == "" || seen[item] {
+			continue
+		}
+		seen[item] = true
+		out = append(out, item)
+	}
+	return out
+}
+
+func buildMultiRepoMarkdown(analysis map[string]interface{}) string {
+	var b strings.Builder
+
+	title := "多仓库协同报告"
+	if t, ok := analysis["title"].(string); ok && t != "" {
+		title = t
+	}
+	b.WriteString("# ")
+	b.WriteString(title)
+	b.WriteString("\n\n")
+
+	// Summary
+	if summary, ok := analysis["summary"].(map[string]interface{}); ok {
+		target := mrString(summary["target_release"])
+		b.WriteString("## 总览\n\n")
+		b.WriteString("| 指标 | 数值 |\n")
+		b.WriteString("|------|------|\n")
+		fmt.Fprintf(&b, "| 仓库数 | %v |\n", summary["repos"])
+		if target != "" {
+			fmt.Fprintf(&b, "| 目标版本 | %s |\n", target)
+		}
+		fmt.Fprintf(&b, "| 开放 Issue | %v |\n", summary["open_issues"])
+		fmt.Fprintf(&b, "| 阻塞 Issue | %v |\n", summary["blocker_issues"])
+		fmt.Fprintf(&b, "| 超 7 天未更新 Issue | %v |\n", summary["stale_issues_7d"])
+		fmt.Fprintf(&b, "| 高优先级 Issue | %v |\n", summary["high_priority"])
+		fmt.Fprintf(&b, "| 开放 PR | %v |\n", summary["open_prs"])
+		fmt.Fprintf(&b, "| 超 3 天未合并 PR | %v |\n", summary["stale_prs_3d"])
+		fmt.Fprintf(&b, "| 冲突 PR | %v |\n", summary["conflict_prs"])
+		fmt.Fprintf(&b, "| 采集错误 | %v |\n\n", summary["collection_errors"])
+	}
+
+	// Issue tracking by repo
+	if it, ok := analysis["issue_tracking"].(map[string]interface{}); ok {
+		b.WriteString("## Issue 追踪\n\n")
+		if rows, ok := it["by_repo"].([]map[string]interface{}); ok && len(rows) > 0 {
+			b.WriteString("| 仓库 | 开放 | 阻塞 | 超7天 | 高优先级 |\n")
+			b.WriteString("|------|------|------|------|----------|\n")
+			for _, row := range rows {
+				fmt.Fprintf(&b, "| %s | %v | %v | %v | %v |\n",
+					mrString(row["repo"]), row["open"], row["blockers"], row["stale_7d"], row["high_priority"])
+			}
+			b.WriteString("\n")
+		}
+		if details, ok := it["details"].([]map[string]interface{}); ok && len(details) > 0 {
+			b.WriteString("### 需关注的 Issue\n\n")
+			b.WriteString("| 仓库 | 编号 | 标题 | 负责人 | 陈旧(天) | 阻塞 | 高优先级 |\n")
+			b.WriteString("|------|------|------|--------|-----------|------|----------|\n")
+			for _, d := range details {
+				blocker := "否"
+				if v, _ := d["blocker"].(bool); v {
+					blocker = "是"
+				}
+				high := "否"
+				if v, _ := d["high"].(bool); v {
+					high = "是"
+				}
+				fmt.Fprintf(&b, "| %s | %v | %s | %s | %v | %s | %s |\n",
+					mrString(d["repo"]), d["number"], mrString(d["title"]),
+					mrString(d["assignee"]), d["stale_days"], blocker, high)
+			}
+			b.WriteString("\n")
+		}
+	}
+
+	// PR board by repo
+	if pr, ok := analysis["pr_board"].(map[string]interface{}); ok {
+		b.WriteString("## PR 看板\n\n")
+		if rows, ok := pr["by_repo"].([]map[string]interface{}); ok && len(rows) > 0 {
+			b.WriteString("| 仓库 | 开放 | 超3天 | 冲突 | 待审查 |\n")
+			b.WriteString("|------|------|------|------|--------|\n")
+			for _, row := range rows {
+				fmt.Fprintf(&b, "| %s | %v | %v | %v | %v |\n",
+					mrString(row["repo"]), row["open"], row["stale_3d"], row["conflicts"], row["needs_review"])
+			}
+			b.WriteString("\n")
+		}
+		if details, ok := pr["details"].([]map[string]interface{}); ok && len(details) > 0 {
+			b.WriteString("### 需关注的 PR\n\n")
+			b.WriteString("| 仓库 | 编号 | 标题 | 负责人 | 陈旧(天) | 冲突 |\n")
+			b.WriteString("|------|------|------|--------|-----------|------|\n")
+			for _, d := range details {
+				conflict := "否"
+				if v, _ := d["conflict"].(bool); v {
+					conflict = "是"
+				}
+				fmt.Fprintf(&b, "| %s | %v | %s | %s | %v | %s |\n",
+					mrString(d["repo"]), d["number"], mrString(d["title"]),
+					mrString(d["assignee"]), d["stale_days"], conflict)
+			}
+			b.WriteString("\n")
+		}
+	}
+
+	// Release coordination
+	if rc, ok := analysis["release_coordination"].(map[string]interface{}); ok {
+		target := mrString(rc["target"])
+		if target != "" {
+			b.WriteString("## Release 协调\n\n")
+			ready := false
+			if v, ok := rc["ready_to_release"].(bool); ok {
+				ready = v
+			}
+			if ready {
+				b.WriteString("**状态：可以发布**\n\n")
+			} else {
+				b.WriteString("**状态：存在阻塞项**\n\n")
+			}
+			if rows, ok := rc["by_repo"].([]map[string]interface{}); ok && len(rows) > 0 {
+				b.WriteString("| 仓库 | 目标版本 | 已发布 | 阻塞Issue | 冲突/陈旧PR |\n")
+				b.WriteString("|------|----------|--------|-----------|-------------|\n")
+				for _, row := range rows {
+					released := "否"
+					if v, _ := row["already_released"].(bool); v {
+						released = "是"
+					}
+					fmt.Fprintf(&b, "| %s | %s | %s | %v | %v |\n",
+						mrString(row["repo"]), mrString(row["target"]), released,
+						row["blocker_issues"], row["stale_or_conflict_pr"])
+				}
+				b.WriteString("\n")
+			}
+			if blockers, ok := rc["blockers"].([]map[string]interface{}); ok && len(blockers) > 0 {
+				b.WriteString("### 阻塞项\n\n")
+				for _, blk := range blockers {
+					fmt.Fprintf(&b, "- %s\n", mrString(blk["reason"]))
+				}
+				b.WriteString("\n")
+			}
+		}
+	}
+
+	// Recommendations
+	if recs, ok := analysis["recommendations"].([]string); ok && len(recs) > 0 {
+		b.WriteString("## 行动建议\n\n")
+		for _, r := range recs {
+			fmt.Fprintf(&b, "- %s\n", r)
+		}
+		b.WriteString("\n")
+	}
+
+	b.WriteString("> 由 multi-repo 工作流自动生成\n")
+	return b.String()
+}

shortcuts/workflow/rules/multi_repo_test.go

@@ -0,0 +1,123 @@
+package rules
+
+import "testing"
+
+func TestMultiRepoCoordinationRuleBlocksRelease(t *testing.T) {
+	upstream := map[string]interface{}{
+		"multi-repo-snapshot": map[string]interface{}{
+			"release": "v1.4.0",
+			"repos": []interface{}{
+				map[string]interface{}{
+					"owner": "org",
+					"repo":  "backend",
+					"open_issues": map[string]interface{}{
+						"issues": []interface{}{
+							map[string]interface{}{
+								"id":          float64(23),
+								"subject":     "登录接口超时 blocker",
+								"priority_id": float64(4),
+								"updated_at":  "2026-06-20T00:00:00Z",
+							},
+						},
+					},
+					"open_prs": map[string]interface{}{
+						"pull_requests": []interface{}{},
+					},
+					"releases": []interface{}{
+						map[string]interface{}{"tag_name": "v1.4.0"},
+					},
+				},
+				map[string]interface{}{
+					"owner": "org",
+					"repo":  "frontend",
+					"open_issues": map[string]interface{}{
+						"issues": []interface{}{},
+					},
+					"open_prs": map[string]interface{}{
+						"pull_requests": []interface{}{
+							map[string]interface{}{
+								"id":           float64(39),
+								"title":        "fix: 修复登录样式 conflict",
+								"updated_at":   "2026-06-25T00:00:00Z",
+								"merge_status": "conflict",
+							},
+						},
+					},
+					"releases": []interface{}{},
+				},
+			},
+		},
+	}
+
+	resp, err := MultiRepoCoordinationRule(upstream, "multi-repo-coordination")
+	if err != nil {
+		t.Fatalf("MultiRepoCoordinationRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	release := analysis["release_coordination"].(map[string]interface{})
+	if release["ready_to_release"].(bool) {
+		t.Fatal("expected release ready_to_release=false")
+	}
+	blockers := release["blockers"].([]map[string]interface{})
+	if len(blockers) != 3 {
+		t.Fatalf("expected blocker issue plus conflict and stale PR blockers, got %v", blockers)
+	}
+	rows := release["by_repo"].([]map[string]interface{})
+	if !rows[0]["already_released"].(bool) {
+		t.Fatalf("expected backend to be marked already_released")
+	}
+	if rows[1]["release_exists"].(bool) {
+		t.Fatalf("expected frontend release_exists=false without treating it as blocker")
+	}
+	summary := analysis["summary"].(map[string]interface{})
+	if summary["blocker_issues"].(int) != 1 {
+		t.Fatalf("blocker_issues = %v, want 1", summary["blocker_issues"])
+	}
+	if summary["conflict_prs"].(int) != 1 {
+		t.Fatalf("conflict_prs = %v, want 1", summary["conflict_prs"])
+	}
+	issues := analysis["issue_tracking"].(map[string]interface{})["details"].([]map[string]interface{})
+	if len(issues) != 1 {
+		t.Fatalf("expected duplicate issue details to be deduped, got %v", issues)
+	}
+}
+
+func TestMultiRepoCoordinationRuleAlreadyReleasedIsNotReadyToRelease(t *testing.T) {
+	upstream := map[string]interface{}{
+		"multi-repo-snapshot": map[string]interface{}{
+			"release": "v1.4.0",
+			"repos": []interface{}{
+				map[string]interface{}{
+					"owner": "org",
+					"repo":  "backend",
+					"open_issues": map[string]interface{}{
+						"issues": []interface{}{},
+					},
+					"open_prs": map[string]interface{}{
+						"pull_requests": []interface{}{},
+					},
+					"releases": []interface{}{
+						map[string]interface{}{"tag_name": "v1.4.0"},
+					},
+				},
+			},
+		},
+	}
+
+	resp, err := MultiRepoCoordinationRule(upstream, "multi-repo-coordination")
+	if err != nil {
+		t.Fatalf("MultiRepoCoordinationRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	release := analysis["release_coordination"].(map[string]interface{})
+	if release["ready_to_release"].(bool) {
+		t.Fatal("already released repo should not be marked ready_to_release")
+	}
+	rows := release["by_repo"].([]map[string]interface{})
+	if !rows[0]["release_exists"].(bool) || !rows[0]["already_released"].(bool) {
+		t.Fatalf("expected release existence flags, got %+v", rows[0])
+	}
+	if rows[0]["ready_to_release"].(bool) {
+		t.Fatalf("already released repo row should not be ready_to_release: %+v", rows[0])
+	}
+}

shortcuts/workflow/rules/registry.go

@@ -0,0 +1,17 @@
+package rules
+
+import "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+
+func init() {
+	workflow.RegisterRuleEngine("gitlink-triage", TriageRule)
+	workflow.RegisterRuleEngine("gitlink-health", HealthDispatchRule)
+	workflow.RegisterRuleEngine("gitlink-contributor-ranking", ContributorRankingRule)
+	workflow.RegisterRuleEngine("gitlink-changelog", ChangelogRule)
+	workflow.RegisterRuleEngine("gitlink-review", CodeReviewRule)
+	workflow.RegisterRuleEngine("gitlink-ci", CIDiagnosisRule)
+	workflow.RegisterRuleEngine("gitlink-license", LicenseCheckRule)
+	workflow.RegisterRuleEngine("gitlink-repo", RepoAuditRule)
+	workflow.RegisterRuleEngine("gitlink-init-scaffold", InitScaffoldRule)
+	workflow.RegisterRuleEngine("gitlink-auto-merge", AutoMergeRule)
+	workflow.RegisterRuleEngine("gitlink-multi-repo", MultiRepoCoordinationRule)
+}

shortcuts/workflow/rules/registry_test.go

@@ -0,0 +1,60 @@
+package rules
+
+import (
+	"testing"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestRegistryAllRegistered(t *testing.T) {
+	expected := []string{
+		"gitlink-triage",
+		"gitlink-health",
+		"gitlink-changelog",
+		"gitlink-review",
+		"gitlink-ci",
+		"gitlink-license",
+		"gitlink-repo",
+		"gitlink-contributor-ranking",
+		"gitlink-init-scaffold",
+		"gitlink-auto-merge",
+		"gitlink-multi-repo",
+	}
+	for _, target := range expected {
+		if _, ok := workflow.RuleEngines[target]; !ok {
+			t.Errorf("rule engine not registered for target %q", target)
+		}
+	}
+	if len(workflow.RuleEngines) != len(expected) {
+		t.Fatalf("expected %d rule engines, got %d", len(expected), len(workflow.RuleEngines))
+	}
+}
+
+func TestHealthDispatchRule(t *testing.T) {
+	// contributor-ranking step should delegate to ContributorRankingRule.
+	resp, err := HealthDispatchRule(map[string]interface{}{}, "contributor-ranking")
+	if err != nil {
+		t.Fatalf("HealthDispatchRule failed: %v", err)
+	}
+	if resp == nil {
+		t.Fatal("expected non-nil response")
+	}
+
+	// health-report step should delegate to HealthReportRule.
+	resp, err = HealthDispatchRule(map[string]interface{}{}, "health-report")
+	if err != nil {
+		t.Fatalf("HealthDispatchRule for health-report failed: %v", err)
+	}
+	if resp == nil {
+		t.Fatal("expected non-nil response for health-report")
+	}
+
+	// repo-health step should delegate to HealthReportRule too.
+	resp, err = HealthDispatchRule(map[string]interface{}{}, "repo-health")
+	if err != nil {
+		t.Fatalf("HealthDispatchRule for repo-health failed: %v", err)
+	}
+	if resp == nil {
+		t.Fatal("expected non-nil response for repo-health")
+	}
+}

shortcuts/workflow/rules/repo_audit.go

@@ -0,0 +1,125 @@
+package rules
+
+import (
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// RepoAuditRule evaluates a repository against a 6-dimension checklist.
+func RepoAuditRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	repoInfo := extractFirst(upstream, "repo-info")
+	labels := extractLabels(upstream)
+	milestones := extractList(upstream, "milestones")
+	branches := extractList(upstream, "branches")
+
+	type dimScore struct {
+		Name   string  `json:"name"`
+		Score  float64 `json:"score"`
+		Weight float64 `json:"weight"`
+		Status string  `json:"status"`
+		Detail string  `json:"detail"`
+	}
+
+	var dims []dimScore
+	missing := []string{}
+
+	// Readme check — use file list since repo +info API doesn't return has_readme.
+	if detectFileInList(upstream, "README.md", "readme.md", "README", "readme") {
+		dims = append(dims, dimScore{Name: "README", Score: 100, Weight: 0.25, Status: "ok", Detail: "README 已存在"})
+	} else {
+		dims = append(dims, dimScore{Name: "README", Score: 0, Weight: 0.25, Status: "missing", Detail: "缺少 README 文件"})
+		missing = append(missing, "README")
+	}
+
+	// License check.
+	if detectLicenseInFiles(upstream) {
+		dims = append(dims, dimScore{Name: "LICENSE", Score: 100, Weight: 0.20, Status: "ok", Detail: "LICENSE 文件已存在"})
+	} else {
+		dims = append(dims, dimScore{Name: "LICENSE", Score: 0, Weight: 0.20, Status: "missing", Detail: "缺少 LICENSE 文件"})
+		missing = append(missing, "LICENSE")
+	}
+
+	// Labels check.
+	if len(labels) > 5 {
+		dims = append(dims, dimScore{Name: "标签库", Score: 100, Weight: 0.15, Status: "ok", Detail: "标签配置完善"})
+	} else if len(labels) > 0 {
+		dims = append(dims, dimScore{Name: "标签库", Score: 50, Weight: 0.15, Status: "partial", Detail: "标签较少，建议补充"})
+	} else {
+		dims = append(dims, dimScore{Name: "标签库", Score: 0, Weight: 0.15, Status: "missing", Detail: "未配置标签"})
+		missing = append(missing, "labels")
+	}
+
+	// Milestones check.
+	if len(milestones) > 0 {
+		dims = append(dims, dimScore{Name: "里程碑", Score: 100, Weight: 0.15, Status: "ok", Detail: "已配置里程碑"})
+	} else {
+		dims = append(dims, dimScore{Name: "里程碑", Score: 0, Weight: 0.15, Status: "missing", Detail: "未配置里程碑"})
+		missing = append(missing, "milestones")
+	}
+
+	// Branches check.
+	if len(branches) > 2 {
+		dims = append(dims, dimScore{Name: "分支结构", Score: 100, Weight: 0.10, Status: "ok", Detail: "分支结构完善"})
+	} else if len(branches) > 1 {
+		dims = append(dims, dimScore{Name: "分支结构", Score: 70, Weight: 0.10, Status: "ok", Detail: "至少有一个开发分支"})
+	} else {
+		dims = append(dims, dimScore{Name: "分支结构", Score: 40, Weight: 0.10, Status: "partial", Detail: "仅主分支，建议创建 develop 分支"})
+	}
+
+	// DevOps check.
+	devops := false
+	if repoInfo != nil {
+		if v, ok := repoInfo["open_devops"].(bool); ok {
+			devops = v
+		}
+		if v, ok := repoInfo["devops_enabled"].(bool); ok {
+			devops = v
+		}
+	}
+	if devops {
+		dims = append(dims, dimScore{Name: "DevOps", Score: 100, Weight: 0.15, Status: "ok", Detail: "DevOps 已开启"})
+	} else {
+		dims = append(dims, dimScore{Name: "DevOps", Score: 0, Weight: 0.15, Status: "missing", Detail: "DevOps 未开启"})
+		missing = append(missing, "DevOps")
+	}
+
+	// Composite.
+	var composite float64
+	for _, d := range dims {
+		composite += d.Score * d.Weight
+	}
+
+	analysis := map[string]interface{}{
+		"composite_score": composite,
+		"grade":           grade(composite),
+		"dimensions":      dims,
+		"missing_items":   missing,
+		"recommendation":  strings.Join(missing, "、") + " 需要补充",
+	}
+	if len(missing) == 0 {
+		analysis["recommendation"] = "项目初始化完善，所有检查项均已通过"
+	}
+
+	return &workflow.AIResponse{Analysis: analysis, Actions: nil}, nil
+}
+
+func detectLicenseInFiles(upstream map[string]interface{}) bool {
+	return detectFileInList(upstream, "LICENSE", "license", "LICENSE.txt", "LICENSE.md", "COPYING")
+}
+
+func detectFileInList(upstream map[string]interface{}, names ...string) bool {
+	files := extractList(upstream, "existing-files")
+	if len(files) == 0 {
+		files = extractList(upstream, "files")
+	}
+	for _, f := range files {
+		name := str(f, "name", "filename", "path", "file_name")
+		for _, n := range names {
+			if strings.EqualFold(name, n) {
+				return true
+			}
+		}
+	}
+	return false
+}

shortcuts/workflow/rules/repo_audit_test.go

@@ -0,0 +1,93 @@
+package rules
+
+import (
+	"testing"
+)
+
+func TestRepoAuditComplete(t *testing.T) {
+	upstream := map[string]interface{}{
+		"repo-info": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{
+					"description":    "A well-maintained project",
+					"has_readme":     true,
+					"open_devops":    true,
+				},
+			},
+		},
+		"existing-files": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"name": "README.md", "content": "# Project"},
+				map[string]interface{}{"name": "LICENSE", "content": "MIT"},
+			},
+		},
+		"labels": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"name": "bug"},
+				map[string]interface{}{"name": "enhancement"},
+				map[string]interface{}{"name": "question"},
+				map[string]interface{}{"name": "docs"},
+				map[string]interface{}{"name": "security"},
+				map[string]interface{}{"name": "refactor"},
+			},
+		},
+		"milestones": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"title": "v1.0"},
+			},
+		},
+		"branches": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"name": "master"},
+				map[string]interface{}{"name": "develop"},
+				map[string]interface{}{"name": "feature/x"},
+			},
+		},
+	}
+
+	resp, err := RepoAuditRule(upstream, "repo-audit")
+	if err != nil {
+		t.Fatalf("RepoAuditRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+
+	// Fully complete repo should have a high score.
+	composite := analysis["composite_score"].(float64)
+	if composite < 80 {
+		t.Errorf("expected composite >= 80 for complete repo, got %.1f", composite)
+	}
+
+	// Should have no missing items.
+	missing := analysis["missing_items"].([]string)
+	if len(missing) != 0 {
+		t.Errorf("expected 0 missing items, got %v", missing)
+	}
+}
+
+func TestRepoAuditEmpty(t *testing.T) {
+	upstream := map[string]interface{}{
+		"repo-info":  map[string]interface{}{"data": []interface{}{}},
+		"labels":     map[string]interface{}{"data": []interface{}{}},
+		"milestones": map[string]interface{}{"data": []interface{}{}},
+		"branches":   map[string]interface{}{"data": []interface{}{}},
+	}
+
+	resp, err := RepoAuditRule(upstream, "repo-audit")
+	if err != nil {
+		t.Fatalf("RepoAuditRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+
+	// Empty repo should have a low score.
+	composite := analysis["composite_score"].(float64)
+	if composite >= 50 {
+		t.Errorf("expected composite < 50 for empty repo, got %.1f", composite)
+	}
+
+	// Should have missing items.
+	missing := analysis["missing_items"].([]string)
+	if len(missing) == 0 {
+		t.Fatal("expected missing items for empty repo")
+	}
+	t.Logf("missing items: %v", missing)
+}

shortcuts/workflow/rules/review.go

@@ -0,0 +1,267 @@
+package rules
+
+import (
+	"fmt"
+	"regexp"
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// finding holds a single code review finding.
+type finding struct {
+	PRNumber interface{} `json:"pr_number"`
+	PRTitle  string      `json:"pr_title"`
+	Lens     string      `json:"lens"`
+	Severity string      `json:"severity"`
+	What     string      `json:"what"`
+	Why      string      `json:"why"`
+	Fix      string      `json:"fix"`
+}
+
+// CodeReviewRule performs static analysis on PR metadata.
+func CodeReviewRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	prs := extractPRs(upstream, "open-prs")
+	if len(prs) == 0 {
+		return &workflow.AIResponse{
+			Analysis: map[string]interface{}{"findings": nil, "message": "no open PRs to review"},
+			Actions:  nil,
+		}, nil
+	}
+
+	var findings []finding
+	var actions []workflow.AIAction
+
+	// Diffs may have been pre-fetched by the workflow engine.
+	prDiffsMap := extractDiffsFromUpstream(upstream)
+
+
+	for _, pr := range prs {
+		title := str(pr, "title", "name")
+		body := str(pr, "body", "description")
+		prNum := interfaceToString(pr["pull_request_number"])
+		if prNum == "" {
+			prNum = interfaceToString(pr["id"])
+			if prNum == "" {
+				prNum = interfaceToString(pr["number"])
+				if prNum == "" {
+					prNum = interfaceToString(pr["pull_request_id"])
+				}
+			}
+		}
+		if prNum == "" {
+			continue
+		}
+
+		diffText := prDiffsMap[prNum]
+		text := title + " " + body + " " + diffText
+
+		var prFindings []finding
+
+		// Security scan.
+		for _, p := range reviewSecurityPatterns {
+			if p.re.MatchString(text) {
+				f := finding{
+					PRNumber: prNum,
+					PRTitle:  title,
+					Lens:     "security",
+					Severity: p.severity,
+					What:     p.what,
+					Why:      "PR 标题/描述中包含可能存在安全风险的代码模式",
+					Fix:      p.fix,
+				}
+				prFindings = append(prFindings, f)
+				findings = append(findings, f)
+			}
+		}
+
+		// Maintainability scan.
+		filesCount := 0
+		if v, ok := pr["files_count"].(float64); ok {
+			filesCount = int(v)
+		}
+		if filesCount > 50 {
+			f := finding{
+				PRNumber: prNum,
+				PRTitle:  title,
+				Lens:     "maintainability",
+				Severity: "medium",
+				What:     fmt.Sprintf("PR 包含 %d 个文件，建议拆分为更小的 PR", filesCount),
+				Why:      "大 PR 难以审查，增加合并风险和回滚难度",
+				Fix:      "将改动按功能模块拆分为多个小 PR",
+			}
+			prFindings = append(prFindings, f)
+			findings = append(findings, f)
+		}
+
+		if body == "" && len(title) < 10 {
+			f := finding{
+				PRNumber: prNum,
+				PRTitle:  title,
+				Lens:     "maintainability",
+				Severity: "low",
+				What:     "PR 缺少描述信息",
+				Why:      "不清晰的 PR 描述增加审查时间，降低代码质量",
+				Fix:      "添加 PR 描述，说明改动原因、影响范围和测试方式",
+			}
+			prFindings = append(prFindings, f)
+			findings = append(findings, f)
+		}
+
+		// Always post a review comment for every PR.
+		commentBody := buildReviewComment(title, prFindings)
+		actions = append(actions, workflow.AIAction{
+			Type:    "cli",
+			Module:  "pr",
+			Command: "+comment",
+			Args: map[string]string{
+				"id":   prNum,
+				"body": commentBody,
+			},
+		})
+
+		// Auto-merge if no issues found.
+		if len(prFindings) == 0 {
+			actions = append(actions, workflow.AIAction{
+				Type:    "cli",
+				Module:  "pr",
+				Command: "+merge",
+				Args: map[string]string{
+					"id":     prNum,
+					"method": "squash",
+				},
+			})
+		}
+	}
+
+	analysis := map[string]interface{}{
+		"reviewed_prs": len(prs),
+		"total_findings": len(findings),
+		"findings":      findings,
+		"summary":       fmt.Sprintf("审查了 %d 个 PR，发现 %d 个问题", len(prs), len(findings)),
+	}
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+type reviewPattern struct {
+	re       *regexp.Regexp
+	severity string
+	what     string
+	fix      string
+}
+
+var reviewSecurityPatterns = []reviewPattern{
+	{
+		regexp.MustCompile(`(?i)(password|passwd|secret|token|api[_-]?key)\s*[:=]\s*['"][^\s'"]{8,}['"]`),
+		"high",
+		"检测到硬编码凭据（密码/Token/密钥）",
+		"将凭据移至环境变量或密钥管理服务，使用占位符替换",
+	},
+	{
+		regexp.MustCompile(`(?i)SELECT\s.*\sFROM\s.*WHERE\s.*\+`),
+		"high",
+		"检测到潜在 SQL 注入模式（字符串拼接构建 SQL）",
+		"使用参数化查询或 ORM 框架",
+	},
+	{
+		regexp.MustCompile(`(?i)innerHTML\s*=|document\.write\(|eval\(`),
+		"medium",
+		"检测到潜在 XSS 风险（innerHTML / eval 使用）",
+		"使用 textContent 替代 innerHTML，避免使用 eval",
+	},
+	{
+		regexp.MustCompile(`(?i)-----BEGIN (RSA |EC |OPENSSH |DSA )?PRIVATE KEY-----`),
+		"high",
+		"检测到私钥明文",
+		"立即删除私钥，使用密钥管理服务",
+	},
+	{
+		regexp.MustCompile(`(?i)ghp_[a-zA-Z0-9]{36}`),
+		"high",
+		"检测到 GitHub 个人访问令牌",
+		"撤销此令牌，使用环境变量存储新令牌",
+	},
+	{
+		regexp.MustCompile(`(?i)os\.system\(|exec\(|subprocess\.call\(`),
+		"medium",
+		"检测到潜在命令注入风险",
+		"避免将用户输入直接拼接到系统命令中，使用参数列表形式",
+	},
+}
+
+// buildReviewComment generates a review comment for a PR.
+func buildReviewComment(prTitle string, prFindings []finding) string {
+	var b strings.Builder
+	b.WriteString("🤖 **代码质量审查报告**\n\n")
+
+	if len(prFindings) == 0 {
+		b.WriteString("✅ **审查通过**：未发现安全风险或代码质量问题，正在自动合并。\n")
+		return b.String()
+	}
+
+	fmt.Fprintf(&b, "审查发现 **%d** 个问题：\n\n", len(prFindings))
+	for _, f := range prFindings {
+		icon := "🔴"
+		switch f.Severity {
+		case "medium":
+			icon = "🟡"
+		case "low":
+			icon = "🟢"
+		}
+		fmt.Fprintf(&b, "- %s **[%s] %s**：%s\n", icon, f.Severity, f.What, f.Why)
+		if f.Fix != "" {
+			fmt.Fprintf(&b, "  - 建议：%s\n", f.Fix)
+		}
+	}
+
+	hasHigh := false
+	for _, f := range prFindings {
+		if f.Severity == "high" {
+			hasHigh = true
+			break
+		}
+	}
+	if hasHigh {
+		b.WriteString("\n⚠️ 存在高危问题，请修复后重新提交审查。\n")
+	} else {
+		b.WriteString("\n请评估以上问题是否需要修复。\n")
+	}
+	return b.String()
+}
+
+func interfaceToString(v interface{}) string {
+	if v == nil {
+		return ""
+	}
+	switch val := v.(type) {
+	case string:
+		return val
+	case float64:
+		return fmt.Sprintf("%.0f", val)
+	case int:
+		return fmt.Sprintf("%d", val)
+	default:
+		return fmt.Sprint(v)
+	}
+}
+
+// extractDiffsFromUpstream retrieves pre-fetched PR diffs from the upstream data.
+func extractDiffsFromUpstream(upstream map[string]interface{}) map[string]string {
+	raw, ok := upstream["_pr_diffs"]
+	if !ok {
+		return nil
+	}
+	if m, ok := raw.(map[string]string); ok {
+		return m
+	}
+	if m, ok := raw.(map[string]interface{}); ok {
+		diffs := make(map[string]string)
+		for k, v := range m {
+			if s, ok := v.(string); ok {
+				diffs[k] = s
+			}
+		}
+		return diffs
+	}
+	return nil
+}

shortcuts/workflow/rules/review_test.go

@@ -0,0 +1,92 @@
+package rules
+
+import (
+	"testing"
+)
+
+func TestCodeReviewStaticAnalysis(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-prs": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{
+					"id":    "1",
+					"title": "add feature",
+					"body":  "password = 'hardcoded12345678'",
+				},
+				map[string]interface{}{
+					"id":          "2",
+					"title":       "wip",
+					"body":        "",
+					"files_count": 60.0,
+				},
+				map[string]interface{}{
+					"id":    "3",
+					"title": "clean refactor with proper description",
+					"body":  "refactoring the auth module to use new token service",
+				},
+			},
+		},
+	}
+
+	resp, err := CodeReviewRule(upstream, "review")
+	if err != nil {
+		t.Fatalf("CodeReviewRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+
+	total := analysis["total_findings"].(int)
+	if total == 0 {
+		t.Fatal("expected findings for hardcoded password and large PR")
+	}
+	findings := analysis["findings"].([]finding)
+	hasSecurity := false
+	hasMaint := false
+	for _, f := range findings {
+		if f.Lens == "security" {
+			hasSecurity = true
+		}
+		if f.Lens == "maintainability" {
+			hasMaint = true
+		}
+	}
+	if !hasSecurity {
+		t.Error("expected security finding for hardcoded password")
+	}
+	if !hasMaint {
+		t.Error("expected maintainability finding for large PR or missing body")
+	}
+
+	// Every PR gets a comment.
+	commentCount := 0
+	mergeCount := 0
+	for _, a := range resp.Actions {
+		if a.Command == "+comment" {
+			commentCount++
+		}
+		if a.Command == "+merge" {
+			mergeCount++
+		}
+	}
+	if commentCount != 3 {
+		t.Fatalf("expected 3 comment actions (one per PR), got %d", commentCount)
+	}
+	// PR #3 has no findings, should be auto-merged.
+	if mergeCount != 1 {
+		t.Fatalf("expected 1 merge action (clean PR), got %d", mergeCount)
+	}
+}
+
+func TestCodeReviewNoPRs(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-prs": map[string]interface{}{"data": []interface{}{}},
+	}
+
+	resp, err := CodeReviewRule(upstream, "review")
+	if err != nil {
+		t.Fatalf("CodeReviewRule failed: %v", err)
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if msg := analysis["message"]; msg != "no open PRs to review" {
+		t.Fatalf("expected 'no open PRs to review', got %v", msg)
+	}
+}

shortcuts/workflow/rules/triage.go

@@ -0,0 +1,355 @@
+package rules
+
+import (
+	"fmt"
+	"regexp"
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// TriageRule classifies issues, assigns priorities, matches labels, and distributes work.
+func TriageRule(upstream map[string]interface{}, stepName string) (*workflow.AIResponse, error) {
+	issues := extractIssues(upstream, "open-issues")
+	labels := extractLabels(upstream)
+	members := extractMembers(upstream)
+
+	if len(issues) == 0 {
+		return &workflow.AIResponse{
+			Analysis: map[string]interface{}{"classified": 0, "message": "no open issues to triage"},
+			Actions:  nil,
+		}, nil
+	}
+
+	var actions []workflow.AIAction
+	memberLoad := map[string]int{}
+	classified := []map[string]interface{}{}
+	gfi := []map[string]interface{}{}
+
+	for _, issue := range issues {
+		num := issueNumber(issue)
+		title := str(issue, "title")
+		body := str(issue, "body", "description")
+		text := title + " " + body
+		open := isIssueOpen(issue)
+		processed := isIssueProcessed(issue)
+
+		cat := classifyIssue(text)
+		pri := assignPriority(text)
+		labelIDs := matchLabels(cat, labels)
+		assignee := leastLoaded(memberLoad, members)
+
+		if assignee != "" {
+			memberLoad[assignee]++
+		}
+
+		result := map[string]interface{}{
+			"number":    num,
+			"title":     title,
+			"category":  cat,
+			"priority":  pri,
+			"assignee":  assignee,
+			"processed": processed,
+		}
+		classified = append(classified, result)
+
+		if processed {
+			continue
+		}
+
+		// Build PATCH action for fields supported by issue update.
+		body2 := map[string]interface{}{}
+		body2["subject"] = title
+		body2["description"] = body
+		if statusID := issueStatusID(issue); statusID != nil {
+			body2["status_id"] = statusID
+		}
+		if len(labelIDs) > 0 {
+			body2["issue_tag_ids"] = labelIDs
+		}
+		if assignee != "" {
+			body2["assigner_ids"] = []string{assignee}
+		}
+		if pri > 0 {
+			body2["priority_id"] = pri
+		}
+		if open && len(body2) > 0 && num != "" {
+			actions = append(actions, workflow.AIAction{
+				Type: "api", Method: "PATCH",
+				Path: fmt.Sprintf("{v1}/issues/%s", num),
+				Body: body2,
+			})
+		}
+
+		if open && isGoodFirstIssue(text) {
+			gfi = append(gfi, result)
+			actions = append(actions, workflow.AIAction{
+				Type: "cli", Module: "issue", Command: "+comment",
+				Args: map[string]string{
+					"number": num,
+					"body":   "👋 感谢提交 Issue！这个 Issue 已被标记为 **Good First Issue**，适合新贡献者参与。欢迎提交 PR！\n\n---\n*此评论由社区运营工作流自动生成*",
+				},
+			})
+		}
+	}
+
+	analysis := map[string]interface{}{
+		"classified":        len(classified),
+		"results":           classified,
+		"good_first_issues": gfi,
+	}
+
+	return &workflow.AIResponse{Analysis: analysis, Actions: actions}, nil
+}
+
+// --- classification ---
+
+var catPatterns = []struct {
+	re       *regexp.Regexp
+	category string
+}{
+	{regexp.MustCompile(`(?i)错误|失败|异常|崩溃|crash|error|bug|broken|404|500`), "bug"},
+	{regexp.MustCompile(`(?i)安全|漏洞|泄露|vulnerability|CVE|敏感`), "security"},
+	{regexp.MustCompile(`(?i)性能|慢|卡顿|优化|performance|speed`), "performance"},
+	{regexp.MustCompile(`(?i)建议|希望|新增|支持|feature|enhancement|add|improve`), "enhancement"},
+	{regexp.MustCompile(`(?i)重构|代码质量|refactor|clean\s*up|tech\s*debt`), "refactor"},
+	{regexp.MustCompile(`(?i)文档|README|帮助|doc|documentation|typo`), "docs"},
+	{regexp.MustCompile(`(?i)如何|怎么|请问|how\s*to|question|help|求助`), "question"},
+}
+
+var priorityPatterns = []struct {
+	re  *regexp.Regexp
+	pri int
+}{
+	{regexp.MustCompile(`(?i)紧急|urgent|critical|崩溃|crash|严重|安全|漏洞|CVE|P0`), 4},
+	{regexp.MustCompile(`(?i)重要|high|important|P1|阻断`), 3},
+	{regexp.MustCompile(`(?i)低|low|trivial|minor|P3`), 1},
+}
+
+func classifyIssue(text string) string {
+	for _, p := range catPatterns {
+		if p.re.MatchString(text) {
+			return p.category
+		}
+	}
+	return "enhancement" // default
+}
+
+func assignPriority(text string) int {
+	for _, p := range priorityPatterns {
+		if p.re.MatchString(text) {
+			return p.pri
+		}
+	}
+	return 2 // default: medium
+}
+
+func isGoodFirstIssue(text string) bool {
+	gfiRe := regexp.MustCompile(`(?i)good\s*first\s*issue|beginner|easy|简单|新手|入门`)
+	if gfiRe.MatchString(text) {
+		return true
+	}
+	// Also mark simple enhancements/docs as GFI.
+	cat := classifyIssue(text)
+	pri := assignPriority(text)
+	return (cat == "docs" || cat == "enhancement") && pri <= 2 &&
+		len(strings.Fields(text)) < 200
+}
+
+func isIssueOpen(issue map[string]interface{}) bool {
+	for _, k := range []string{"state", "status", "issue_status", "status_name"} {
+		s := strings.ToLower(strings.TrimSpace(str(issue, k)))
+		switch s {
+		case "closed", "close", "resolved", "done", "已关闭", "关闭", "已解决":
+			return false
+		case "open", "opened", "active", "new", "新增", "开启", "打开":
+			return true
+		}
+	}
+	for _, k := range []string{"status_id", "state_id"} {
+		switch fmt.Sprint(issue[k]) {
+		case "3", "5":
+			return false
+		case "1", "2":
+			return true
+		}
+	}
+	return true
+}
+
+func issueStatusID(issue map[string]interface{}) interface{} {
+	for _, k := range []string{"status_id", "state_id"} {
+		if v := issue[k]; v != nil {
+			return v
+		}
+	}
+	return nil
+}
+
+func isIssueProcessed(issue map[string]interface{}) bool {
+	return commentCount(issue) > 0 || len(extractIssueTags(issue)) > 0 || hasAssignee(issue)
+}
+
+func hasAssignee(issue map[string]interface{}) bool {
+	for _, k := range []string{"assigners", "assignees", "assigned_to", "assignee", "assigned_to_id", "assigner_ids"} {
+		v, ok := issue[k]
+		if !ok || v == nil {
+			continue
+		}
+		switch x := v.(type) {
+		case []interface{}:
+			if len(x) > 0 {
+				return true
+			}
+		case []string:
+			if len(x) > 0 {
+				return true
+			}
+		case string:
+			if strings.TrimSpace(x) != "" {
+				return true
+			}
+		default:
+			if fmt.Sprint(x) != "" && fmt.Sprint(x) != "0" && fmt.Sprint(x) != "<nil>" {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+// --- label matching ---
+
+func extractLabels(upstream map[string]interface{}) []map[string]interface{} {
+	return extractList(upstream, "labels")
+}
+
+func matchLabels(category string, labels []map[string]interface{}) []interface{} {
+	catLower := strings.ToLower(category)
+	aliases := labelAliases(catLower)
+	var ids []interface{}
+	for _, l := range labels {
+		name := strings.ToLower(str(l, "name", "title", "label"))
+		if name == "" {
+			continue
+		}
+		// Direct match or contains.
+		if matchesLabelName(catLower, aliases, name) {
+			if id := labelID(l); id != nil {
+				ids = append(ids, id)
+			}
+		}
+	}
+	// Also match sub-categories for bug.
+	if catLower == "bug" {
+		for _, l := range labels {
+			name := strings.ToLower(str(l, "name", "title", "label"))
+			if strings.Contains(name, "bug") || strings.Contains(name, "fix") {
+				if id := labelID(l); id != nil {
+					ids = append(ids, id)
+				}
+			}
+		}
+	}
+	return ids
+}
+
+func labelAliases(category string) []string {
+	switch category {
+	case "bug":
+		return []string{"bug", "bugs", "fix", "修复", "疑修", "缺陷", "错误", "故障", "问题"}
+	case "security":
+		return []string{"security", "安全", "漏洞", "cve"}
+	case "performance":
+		return []string{"performance", "perf", "性能", "优化"}
+	case "refactor":
+		return []string{"refactor", "重构", "代码质量", "技术债"}
+	case "enhancement":
+		return []string{"enhancement", "feature", "功能", "需求", "新增", "改进"}
+	case "docs":
+		return []string{"docs", "documentation", "文档", "readme", "帮助"}
+	case "question":
+		return []string{"question", "help", "疑问", "问题", "求助"}
+	default:
+		return []string{category}
+	}
+}
+
+func matchesLabelName(category string, aliases []string, name string) bool {
+	if name == category || strings.Contains(name, category) || strings.Contains(category, name) {
+		return true
+	}
+	for _, alias := range aliases {
+		alias = strings.ToLower(alias)
+		if name == alias || strings.Contains(name, alias) || strings.Contains(alias, name) {
+			return true
+		}
+	}
+	return false
+}
+
+func labelID(l map[string]interface{}) interface{} {
+	for _, k := range []string{"id", "tag_id", "label_id"} {
+		if v := l[k]; v != nil {
+			return v
+		}
+	}
+	return nil
+}
+
+// --- assignment ---
+
+func leastLoaded(load map[string]int, members map[string]string) string {
+	if len(members) == 0 {
+		return ""
+	}
+	best := ""
+	bestN := -1
+	for login, id := range members {
+		n := load[login]
+		if bestN < 0 || n < bestN {
+			bestN = n
+			best = id
+		}
+	}
+	return best
+}
+
+// --- helpers ---
+
+func issueNumber(issue map[string]interface{}) string {
+	for _, k := range []string{"project_issues_index", "number", "iid", "id"} {
+		s := fmt.Sprint(issue[k])
+		if s != "" && s != "0" && s != "<nil>" {
+			return s
+		}
+	}
+	return ""
+}
+
+// extractIssueTags returns the existing tags on an issue.
+func extractIssueTags(issue map[string]interface{}) []interface{} {
+	for _, k := range []string{"issue_tags", "tags", "labels"} {
+		if v, ok := issue[k]; ok {
+			if arr, ok := v.([]interface{}); ok && len(arr) > 0 {
+				return arr
+			}
+		}
+	}
+	return nil
+}
+
+// commentCount returns the number of comments on an issue.
+func commentCount(issue map[string]interface{}) int {
+	for _, k := range []string{"comment_journals_count", "comments_count", "comment_count"} {
+		if v, ok := issue[k]; ok {
+			switch n := v.(type) {
+			case float64:
+				return int(n)
+			case int:
+				return n
+			}
+		}
+	}
+	return 0
+}

shortcuts/workflow/rules/triage_test.go

@@ -0,0 +1,253 @@
+package rules
+
+import (
+	"testing"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestTriageRuleClassifiesByKeyword(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-issues": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"project_issues_index": "1", "title": "fix crash on startup", "body": "应用启动时崩溃"},
+				map[string]interface{}{"project_issues_index": "2", "title": "新增导出功能", "body": ""},
+				map[string]interface{}{"project_issues_index": "3", "title": "如何配置SSO", "body": "请问怎么配"},
+			},
+		},
+		"labels": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": 1, "name": "bug"},
+				map[string]interface{}{"id": 2, "name": "enhancement"},
+				map[string]interface{}{"id": 3, "name": "question"},
+			},
+		},
+		"members": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"login": "dev1", "name": "Dev One"},
+			},
+		},
+	}
+
+	resp, err := TriageRule(upstream, "triage")
+	if err != nil {
+		t.Fatalf("TriageRule failed: %v", err)
+	}
+	if resp.Analysis == nil {
+		t.Fatal("expected non-nil Analysis")
+	}
+
+	analysis, ok := resp.Analysis.(map[string]interface{})
+	if !ok {
+		t.Fatal("Analysis is not a map")
+	}
+	if v := analysis["classified"]; v.(int) != 3 {
+		t.Fatalf("expected 3 classified, got %v", v)
+	}
+	if len(resp.Actions) == 0 {
+		t.Fatal("expected actions for issue triage")
+	}
+
+	// Verify each action type.
+	for _, a := range resp.Actions {
+		if a.Type == "api" && a.Method != "PATCH" {
+			t.Errorf("unexpected API method: %s", a.Method)
+		}
+	}
+}
+
+func TestTriageRuleEmptyIssues(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-issues": map[string]interface{}{"data": []interface{}{}},
+		"labels":      map[string]interface{}{"data": []interface{}{}},
+		"members":     map[string]interface{}{"data": []interface{}{}},
+	}
+	resp, err := TriageRule(upstream, "triage")
+	if err != nil {
+		t.Fatalf("TriageRule failed: %v", err)
+	}
+	if len(resp.Actions) != 0 {
+		t.Fatalf("expected 0 actions for empty issues, got %d", len(resp.Actions))
+	}
+	analysis := resp.Analysis.(map[string]interface{})
+	if v := analysis["classified"]; v.(int) != 0 {
+		t.Fatalf("expected 0 classified, got %v", v)
+	}
+}
+
+func TestTriageRuleGoodFirstIssue(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-issues": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"project_issues_index": "10", "title": "good first issue: add docs", "body": "easy task for beginners"},
+			},
+		},
+		"labels": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": 1, "name": "docs"},
+			},
+		},
+		"members": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"login": "dev1", "name": "Dev"},
+			},
+		},
+	}
+
+	resp, err := TriageRule(upstream, "triage")
+	if err != nil {
+		t.Fatalf("TriageRule failed: %v", err)
+	}
+	hasComment := false
+	for _, a := range resp.Actions {
+		if a.Type == "cli" && a.Module == "issue" && a.Command == "+comment" {
+			hasComment = true
+			break
+		}
+	}
+	if !hasComment {
+		t.Fatal("expected a cli comment action for good first issue")
+	}
+}
+
+func TestTriageRuleLabelsWithChineseAlias(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-issues": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"project_issues_index": "21", "title": "新增代码质量看门人工作流", "body": ""},
+			},
+		},
+		"labels": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": 323830, "name": "功能"},
+			},
+		},
+		"members": map[string]interface{}{"data": []interface{}{}},
+	}
+
+	resp, err := TriageRule(upstream, "triage")
+	if err != nil {
+		t.Fatalf("TriageRule failed: %v", err)
+	}
+	for _, a := range resp.Actions {
+		if a.Type == "api" {
+			ids, ok := a.Body["issue_tag_ids"].([]interface{})
+			if !ok || len(ids) != 1 || ids[0] != 323830 {
+				t.Fatalf("unexpected issue_tag_ids: %+v", a.Body["issue_tag_ids"])
+			}
+			return
+		}
+	}
+	t.Fatal("expected PATCH action with Chinese 功能 label")
+}
+
+func TestTriageRuleDoesNotCommentClosedIssue(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-issues": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"project_issues_index": "10", "title": "good first issue: add docs", "body": "easy task for beginners", "status": "closed"},
+			},
+		},
+		"labels": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": 1, "name": "docs"},
+			},
+		},
+		"members": map[string]interface{}{"data": []interface{}{}},
+	}
+
+	resp, err := TriageRule(upstream, "triage")
+	if err != nil {
+		t.Fatalf("TriageRule failed: %v", err)
+	}
+	for _, a := range resp.Actions {
+		if a.Type == "cli" && a.Module == "issue" && a.Command == "+comment" {
+			t.Fatal("did not expect a cli comment action for closed issue")
+		}
+		if a.Type == "api" && a.Method == "PATCH" {
+			t.Fatal("did not expect a patch action for closed issue")
+		}
+	}
+}
+
+func TestTriageRuleSkipsProcessedIssue(t *testing.T) {
+	upstream := map[string]interface{}{
+		"open-issues": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{
+					"project_issues_index":   "11",
+					"title":                  "新增导出功能",
+					"body":                   "简单功能",
+					"comment_journals_count": 1,
+					"tags":                   []interface{}{map[string]interface{}{"id": 323830, "name": "功能"}},
+					"assigners":              []interface{}{map[string]interface{}{"id": 148915, "login": "yetja"}},
+				},
+			},
+		},
+		"labels": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": 323830, "name": "功能"},
+			},
+		},
+		"members": map[string]interface{}{
+			"data": []interface{}{
+				map[string]interface{}{"id": 148915, "login": "yetja"},
+			},
+		},
+	}
+
+	resp, err := TriageRule(upstream, "triage")
+	if err != nil {
+		t.Fatalf("TriageRule failed: %v", err)
+	}
+	if len(resp.Actions) != 0 {
+		t.Fatalf("expected processed issue to be ignored, got actions: %+v", resp.Actions)
+	}
+}
+
+func TestTriageRulePriority(t *testing.T) {
+	cases := []struct {
+		title    string
+		expected int
+	}{
+		{"紧急: 安全漏洞", 4},
+		{"重要功能", 3},
+		{"普通建议", 2},
+		{"低优先级改进", 1},
+	}
+	for _, tc := range cases {
+		upstream := map[string]interface{}{
+			"open-issues": map[string]interface{}{
+				"data": []interface{}{
+					map[string]interface{}{"project_issues_index": "1", "title": tc.title, "body": ""},
+				},
+			},
+			"labels":  map[string]interface{}{"data": []interface{}{}},
+			"members": map[string]interface{}{"data": []interface{}{}},
+		}
+		resp, err := TriageRule(upstream, "triage")
+		if err != nil {
+			t.Fatalf("TriageRule failed for %q: %v", tc.title, err)
+		}
+		if len(resp.Actions) > 0 {
+			body := resp.Actions[0].Body
+			if v, ok := body["priority_id"]; ok {
+				if v.(int) != tc.expected {
+					t.Errorf("title=%q: priority_id=%v, want %d", tc.title, v, tc.expected)
+				}
+			}
+		}
+	}
+}
+
+func TestTriageRuleOutputFormat(t *testing.T) {
+	resp, err := TriageRule(map[string]interface{}{}, "triage")
+	if err != nil {
+		t.Fatalf("TriageRule failed: %v", err)
+	}
+	if resp == nil {
+		t.Fatal("expected non-nil response")
+	}
+	// Verify it's a valid workflow.AIResponse.
+	var _ *workflow.AIResponse = resp
+}

shortcuts/workflow/skills/gitlink-auto-merge/SKILL.md

@@ -0,0 +1,36 @@
+# gitlink-auto-merge（质量自动合并）
+
+基于 Review 和 CI 诊断结果，当代码质量达标时自动合并 PR。
+
+## 合并条件
+
+1. Review 中不存在 high-severity 的 security 类型发现问题
+2. CI 构建健康（无失败构建，或仓库未配置 CI）
+
+## 操作
+
+当条件满足时，为每个符合条件的 PR 执行：
+
+```
+pr +merge --id <number> --method squash
+```
+
+## 条件不满足时
+
+不执行合并，返回原因：
+- "Review 发现高危安全问题，阻止自动合并"
+- "CI 构建未通过，阻止自动合并"
+
+## 输出格式
+
+```json
+{
+  "analysis": {
+    "merged": 1,
+    "reason": "质量达标，已合并 1 个 PR"
+  },
+  "actions": [
+    {"type": "cli", "module": "pr", "command": "+merge", "args": {"id": "6", "method": "squash"}}
+  ]
+}
+```

shortcuts/workflow/skills/gitlink-contributor-ranking/SKILL.md

@@ -0,0 +1,85 @@
+---
+name: gitlink-contributor-ranking
+version: 1.0.0
+description: "贡献者排行与成长体系：统计贡献者活跃度、生成排行榜、识别新星与流失风险，并自动颁发成就徽章。"
+metadata:
+  requires:
+    bins: ["gitlink-cli"]
+  cliHelp: "gitlink-cli --help"
+---
+
+# gitlink-contributor-ranking（贡献者成长体系）
+
+**CRITICAL — 开始前必须先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)，其中包含认证、权限处理和 API 注意事项。**
+**CRITICAL — GitLink 操作只能用 `gitlink-cli`。禁止用 `gh`（GitHub CLI）操作 GitLink 资源。**
+
+## 概述
+
+本 Skill 用于分析项目贡献者数据，生成贡献者排行榜，识别社区新星和流失风险，并为符合条件的贡献者颁发成就徽章。
+
+## 输入数据
+
+上游已采集以下数据（通过 JSON 传入）：
+
+- `commits` — 最近 50-100 条提交记录
+- `open-issues` — 开放中的 Issue 列表
+- `closed-issues` — 已关闭的 Issue 列表
+- `merged-prs` — 已合并的 PR 列表
+- `members` — 项目成员/协作者列表
+
+每条记录包含作者信息、时间戳等元数据。
+
+## 输出格式
+
+**CRITICAL — 你只需要输出贡献者排行榜和徽章建议，不需要生成项目健康度报告（那是社区运营工作流的事）。**
+
+```markdown
+# 贡献者排行榜
+
+> 统计周期：最近 30 天 | 生成时间：YYYY-MM-DD HH:mm
+
+## 总排行
+
+| 排名 | 贡献者 | 提交 | Issue | PR | 总计 | 趋势 | 勋章 |
+|------|--------|------|-------|-----|------|------|------|
+| 1 | ... | N | N | N | N | ↑/↓/→ | 🥇 |
+| 2 | ... | N | N | N | N | ↑/↓/→ | 🥈 |
+
+## 新星 🌟
+
+- **贡献者名** — 近期活跃度显著提升（趋势 > 50%），建议颁发「新星」徽章
+
+## 流失风险 ⚠️
+
+- **贡献者名** — 超过 30 天无活动记录，建议社区管理员关注
+
+## 徽章颁发建议
+
+根据分析结果，建议为以下贡献者颁发徽章：
+
+| 贡献者 | 建议徽章 | 原因 |
+|--------|----------|------|
+| xxx | 代码贡献者 | 近 30 天提交 N 次 |
+| yyy | Issue 猎手 | 关闭 N 个 Issue |
+| zzz | 新星 | 活跃度上升 N% |
+
+> 由 contributor-growth 工作流自动生成
+```
+
+## 徽章规则
+
+根据数据自动判定：
+
+| 条件 | 徽章 |
+|------|------|
+| 近 30 天提交 ≥ 10 次 | 代码贡献者 |
+| 近 30 天关闭 Issue ≥ 5 个 | Issue 猎手 |
+| 近 30 天合并 PR ≥ 3 个 | PR 达人 |
+| 趋势上升 > 50% | 新星 |
+
+## 注意事项
+
+- 输出纯 Markdown，不要包含 JSON 包装
+- 如果某项数据为空（如无 PR），跳过对应徽章
+- 排行至少展示 Top 10，如果总数不足则全部展示
+- 趋势：近 30 天 vs 前 30 天对比，超过 30 天无活动标注「流失风险」

shortcuts/workflow/skills/gitlink-init-scaffold/SKILL.md

@@ -0,0 +1,36 @@
+# gitlink-init-scaffold
+
+根据项目描述一键创建仓库并初始化脚手架。
+
+## 输入
+
+上游数据（`_desc`）包含项目的自然语言描述。
+
+## 输出
+
+```json
+{
+  "analysis": {
+    "repo": "owner/repo-name",
+    "description": "项目描述",
+    "created": true
+  },
+  "actions": [
+    {"type": "cli", "module": "repo", "command": "+create", "args": {"name": "repo-name", "description": "..."}},
+    {"type": "cli", "module": "file", "command": "+create", "args": {"path": "README.md", "content": "..."}},
+    {"type": "cli", "module": "file", "command": "+create", "args": {"path": "LICENSE", "content": "MIT"}},
+    {"type": "cli", "module": "file", "command": "+create", "args": {"path": ".gitignore", "content": "..."}},
+    {"type": "cli", "module": "label", "command": "+create", "args": {"name": "bug", "color": "#d73a4a"}},
+    {"type": "cli", "module": "milestone", "command": "+create", "args": {"title": "v0.1.0"}},
+    {"type": "cli", "module": "issue", "command": "+create", "args": {"title": "项目初始化", "body": "..."}}
+  ]
+}
+```
+
+## 规则
+
+1. 从 `_desc` 提取英文关键词生成仓库名；若无英文词则用 `_repo` 字段
+2. 创建 README（项目名 + 描述 + 快速开始）、MIT LICENSE、Go .gitignore
+3. 创建 7 个默认标签：bug/security/performance/enhancement/refactor/docs/question
+4. 创建 v0.1.0 里程碑
+5. 创建 3 个初始 Issue：项目初始化、CI/CD 配置、文档完善

shortcuts/workflow/skills/gitlink-multi-repo/SKILL.md

@@ -0,0 +1,26 @@
+# gitlink-multi-repo
+
+多仓库协同工作流：基于多个仓库的 Issue、PR、Release 和 Milestone 快照，生成统一 Issue 追踪、PR 状态看板和 Release 协调发布报告。
+
+## 原则
+
+- 核心统计和发布阻塞判断以规则引擎结果为准。
+- AI 只做语义增强：总结风险、解释阻塞原因、生成行动清单和 Markdown 报告。
+- 不执行写操作，不自动创建 Release，不批量评论 Issue/PR。
+
+## 输入
+
+上游数据来自 `multi-repo-snapshot`，结构包含：
+
+- `release`: 目标版本，例如 `v1.4.0`
+- `repos`: 每个仓库的 `info`、`open_issues`、`open_prs`、`releases`、`milestones`
+- `errors`: 采集失败项
+
+## 输出建议
+
+输出应包含：
+
+- 统一 Issue 追踪：按仓库统计 open、blocker、超期、高优先级 Issue。
+- PR 状态看板：按仓库统计 open、待 review、冲突、超期 PR。
+- Release 协调：判断目标版本是否建议发布，列出阻塞原因。
+- 行动建议：按优先级列出需要处理的仓库和事项。

shortcuts/workflow/snapshot/snapshot.go

@@ -0,0 +1,221 @@
+package snapshot
+
+import (
+	"encoding/csv"
+	"fmt"
+	"net/url"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+)
+
+// MultiRepoSnapshot is a cross-repository state snapshot.
+type MultiRepoSnapshot struct {
+	Release   string         `json:"release,omitempty"`
+	Repos     []RepoSnapshot `json:"repos"`
+	Errors    []RepoError    `json:"errors,omitempty"`
+	Generated string         `json:"generated_at"`
+}
+
+// RepoSnapshot holds collected data for a single repository.
+type RepoSnapshot struct {
+	Owner      string      `json:"owner"`
+	Repo       string      `json:"repo"`
+	Info       interface{} `json:"info,omitempty"`
+	OpenIssues interface{} `json:"open_issues,omitempty"`
+	OpenPRs    interface{} `json:"open_prs,omitempty"`
+	Releases   interface{} `json:"releases,omitempty"`
+	Milestones interface{} `json:"milestones,omitempty"`
+}
+
+// RepoError records an error encountered while collecting repo data.
+type RepoError struct {
+	Owner string `json:"owner"`
+	Repo  string `json:"repo"`
+	Step  string `json:"step"`
+	Error string `json:"error"`
+}
+
+type repoRef = RepoRef
+
+// BuildMultiRepoSnapshot collects Issue/PR/Release/Milestone data for multiple repos.
+func BuildMultiRepoSnapshot(ctx *common.RuntimeContext) (*MultiRepoSnapshot, error) {
+	repos, err := parseMultiRepoRefs(ctx.Arg("repos"), ctx.Arg("from"))
+	if err != nil {
+		return nil, err
+	}
+	if len(repos) == 0 {
+		return nil, fmt.Errorf("multi-repo workflow requires --repos owner/repo[,owner/repo...] or --from repos.csv")
+	}
+
+	snapshot := &MultiRepoSnapshot{
+		Release:   ctx.Arg("release"),
+		Repos:     make([]RepoSnapshot, 0, len(repos)),
+		Generated: time.Now().Format(time.RFC3339),
+	}
+
+	for _, ref := range repos {
+		repoSnap := RepoSnapshot{Owner: ref.Owner, Repo: ref.Repo}
+		collectRepoSnapshot(ctx, ref, &repoSnap, &snapshot.Errors)
+		snapshot.Repos = append(snapshot.Repos, repoSnap)
+	}
+
+	return snapshot, nil
+}
+
+func collectRepoSnapshot(ctx *common.RuntimeContext, ref repoRef, snap *RepoSnapshot, errs *[]RepoError) {
+	call := func(step, method, path string, q url.Values) interface{} {
+		env, err := ctx.CallAPIWithQuery(method, path, q)
+		if err != nil {
+			*errs = append(*errs, RepoError{Owner: ref.Owner, Repo: ref.Repo, Step: step, Error: err.Error()})
+			return nil
+		}
+		if env == nil || !env.OK {
+			msg := "request failed"
+			if env != nil && env.Error != nil {
+				msg = env.Error.Message
+			}
+			*errs = append(*errs, RepoError{Owner: ref.Owner, Repo: ref.Repo, Step: step, Error: msg})
+			return nil
+		}
+		return env.Data
+	}
+
+	base := fmt.Sprintf("/%s/%s", ref.Owner, ref.Repo)
+	v1 := fmt.Sprintf("/v1/%s/%s", ref.Owner, ref.Repo)
+
+	snap.Info = call("repo-info", "GET", base, nil)
+
+	issueQ := url.Values{}
+	issueQ.Set("page", "1")
+	issueQ.Set("limit", "100")
+	issueQ.Set("state", "open")
+	snap.OpenIssues = call("open-issues", "GET", v1+"/issues", issueQ)
+
+	prQ := url.Values{}
+	prQ.Set("page", "1")
+	prQ.Set("limit", "100")
+	prQ.Set("state", "open")
+	snap.OpenPRs = call("open-prs", "GET", base+"/pulls", prQ)
+
+	releaseQ := url.Values{}
+	releaseQ.Set("page", "1")
+	releaseQ.Set("limit", "100")
+	snap.Releases = call("releases", "GET", base+"/releases", releaseQ)
+
+	milestoneQ := url.Values{}
+	milestoneQ.Set("page", "1")
+	milestoneQ.Set("limit", "100")
+	milestoneQ.Set("category", "opening")
+	milestoneQ.Set("sort_by", "created_on")
+	milestoneQ.Set("sort_direction", "desc")
+	snap.Milestones = call("milestones", "GET", v1+"/milestones", milestoneQ)
+}
+
+// ParseRepoCSV reads repo references from a CSV file.
+func ParseRepoCSV(path string) ([]RepoRef, error) {
+	refs, err := parseRepoCSV(path)
+	if err != nil {
+		return nil, err
+	}
+	out := make([]RepoRef, len(refs))
+	for i, r := range refs {
+		out[i] = RepoRef(r)
+	}
+	return out, nil
+}
+
+// ParseMultiRepoRefs parses repo references from a comma-separated string and/or CSV file.
+func ParseMultiRepoRefs(reposArg, fromPath string) ([]RepoRef, error) {
+	return parseMultiRepoRefs(reposArg, fromPath)
+}
+
+// RepoRef is an owner/repo pair.
+type RepoRef struct {
+	Owner string
+	Repo  string
+}
+
+func parseMultiRepoRefs(reposArg, fromPath string) ([]repoRef, error) {
+	var refs []repoRef
+	if reposArg != "" {
+		for _, raw := range strings.Split(reposArg, ",") {
+			ref, err := parseRepoRef(raw)
+			if err != nil {
+				return nil, err
+			}
+			refs = append(refs, ref)
+		}
+	}
+	if fromPath != "" {
+		fileRefs, err := parseRepoCSV(fromPath)
+		if err != nil {
+			return nil, err
+		}
+		refs = append(refs, fileRefs...)
+	}
+	return dedupeRepoRefs(refs), nil
+}
+
+func parseRepoRef(raw string) (repoRef, error) {
+	raw = strings.TrimSpace(raw)
+	parts := strings.Split(raw, "/")
+	if len(parts) != 2 || strings.TrimSpace(parts[0]) == "" || strings.TrimSpace(parts[1]) == "" {
+		return repoRef{}, fmt.Errorf("invalid repo %q: expected owner/repo", raw)
+	}
+	return repoRef{Owner: strings.TrimSpace(parts[0]), Repo: strings.TrimSpace(parts[1])}, nil
+}
+
+func parseRepoCSV(path string) ([]repoRef, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return nil, fmt.Errorf("read repo file %s: %w", path, err)
+	}
+	defer f.Close()
+
+	reader := csv.NewReader(f)
+	reader.FieldsPerRecord = -1
+	records, err := reader.ReadAll()
+	if err != nil {
+		return nil, fmt.Errorf("parse repo file %s: %w", path, err)
+	}
+
+	var refs []repoRef
+	for i, rec := range records {
+		if len(rec) == 0 {
+			continue
+		}
+		if i == 0 && len(rec) >= 2 && strings.EqualFold(strings.TrimSpace(rec[0]), "owner") && strings.EqualFold(strings.TrimSpace(rec[1]), "repo") {
+			continue
+		}
+		if len(rec) == 1 {
+			ref, err := parseRepoRef(rec[0])
+			if err != nil {
+				return nil, fmt.Errorf("line %d: %w", i+1, err)
+			}
+			refs = append(refs, ref)
+			continue
+		}
+		if strings.TrimSpace(rec[0]) == "" || strings.TrimSpace(rec[1]) == "" {
+			return nil, fmt.Errorf("line %d: owner and repo must not be empty", i+1)
+		}
+		refs = append(refs, repoRef{Owner: strings.TrimSpace(rec[0]), Repo: strings.TrimSpace(rec[1])})
+	}
+	return refs, nil
+}
+
+func dedupeRepoRefs(refs []repoRef) []repoRef {
+	seen := make(map[string]bool, len(refs))
+	out := make([]repoRef, 0, len(refs))
+	for _, ref := range refs {
+		key := ref.Owner + "/" + ref.Repo
+		if seen[key] {
+			continue
+		}
+		seen[key] = true
+		out = append(out, ref)
+	}
+	return out
+}

shortcuts/workflow/snapshot/snapshot_test.go

@@ -0,0 +1,38 @@
+package snapshot
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+func TestParseMultiRepoRefs(t *testing.T) {
+	refs, err := ParseMultiRepoRefs("org/backend, org/frontend,org/backend", "")
+	if err != nil {
+		t.Fatalf("ParseMultiRepoRefs failed: %v", err)
+	}
+	if len(refs) != 2 {
+		t.Fatalf("expected 2 deduped refs, got %d: %+v", len(refs), refs)
+	}
+	if refs[0].Owner != "org" || refs[0].Repo != "backend" {
+		t.Fatalf("unexpected first ref: %+v", refs[0])
+	}
+}
+
+func TestParseRepoCSV(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "repos.csv")
+	if err := os.WriteFile(path, []byte("owner,repo\norg,backend\norg/frontend\n"), 0600); err != nil {
+		t.Fatal(err)
+	}
+	refs, err := ParseRepoCSV(path)
+	if err != nil {
+		t.Fatalf("ParseRepoCSV failed: %v", err)
+	}
+	if len(refs) != 2 {
+		t.Fatalf("expected 2 refs, got %d", len(refs))
+	}
+	if refs[1].Owner != "org" || refs[1].Repo != "frontend" {
+		t.Fatalf("unexpected second ref: %+v", refs[1])
+	}
+}

shortcuts/workflow/state/review.go

@@ -0,0 +1,231 @@
+package state
+
+import (
+	"crypto/md5"
+	"encoding/hex"
+	"encoding/json"
+	"fmt"
+	"os"
+	"os/exec"
+	"strings"
+
+	"github.com/gitlink-org/gitlink-cli/shortcuts/common"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// FilterReviewedPRs removes PRs from upstream that were already reviewed
+// with the same fingerprint, preventing re-review on every poll cycle.
+func FilterReviewedPRs(upstream map[string]interface{}, ctx *common.RuntimeContext) {
+	wfName := ctx.Arg("__wf_name")
+	if wfName == "" {
+		return
+	}
+	state, err := LoadState(wfName)
+	if err != nil || state == nil {
+		return
+	}
+	if state.ReviewedPRs == nil {
+		state.ReviewedPRs = make(map[string]string)
+	}
+
+	prs := extractPRListFromUpstream(upstream, "open-prs")
+	if len(prs) == 0 {
+		return
+	}
+
+	filtered := make([]interface{}, 0, len(prs))
+	skipped := 0
+	for _, pr := range prs {
+		prNum := prNumberFromMap(pr)
+		if prNum == "" {
+			filtered = append(filtered, pr)
+			continue
+		}
+		fp := prFingerprint(pr)
+		if stored, ok := state.ReviewedPRs[prNum]; ok && stored == fp {
+			skipped++
+			continue
+		}
+		filtered = append(filtered, pr)
+	}
+
+	if skipped > 0 {
+		fmt.Fprintf(os.Stderr, "[workflow] 跳过 %d 个已审查的 PR（无变化）\n", skipped)
+	}
+
+	upstream["open-prs"] = map[string]interface{}{"data": filtered}
+}
+
+// SaveReviewedPRFingerprints stores PR fingerprints after a successful review.
+func SaveReviewedPRFingerprints(upstream map[string]interface{}, ctx *common.RuntimeContext) {
+	wfName := ctx.Arg("__wf_name")
+	if wfName == "" {
+		return
+	}
+	state, err := LoadState(wfName)
+	if err != nil || state == nil {
+		return
+	}
+	if state.ReviewedPRs == nil {
+		state.ReviewedPRs = make(map[string]string)
+	}
+
+	prs := extractPRListFromUpstream(upstream, "open-prs")
+	for _, pr := range prs {
+		prNum := prNumberFromMap(pr)
+		if prNum == "" {
+			continue
+		}
+		state.ReviewedPRs[prNum] = prFingerprint(pr)
+	}
+	state.Save()
+}
+
+// EnrichWithPRDiffs fetches diffs for open PRs and stores them in upstream.
+func EnrichWithPRDiffs(upstream map[string]interface{}, owner, repo string) {
+	prs := extractPRListFromUpstream(upstream, "open-prs")
+	if len(prs) == 0 {
+		return
+	}
+	diffs := make(map[string]string)
+	for _, pr := range prs {
+		status := ""
+		switch v := pr["pull_request_status"].(type) {
+		case string:
+			status = v
+		case float64:
+			if v == 0 {
+				status = "open"
+			}
+		}
+		if status == "" {
+			if s, ok := pr["pull_request_staus"].(string); ok {
+				status = s
+			}
+		}
+		if status != "" && status != "open" {
+			continue
+		}
+		prNum := ""
+		if n, ok := pr["pull_request_number"]; ok {
+			prNum = fmt.Sprintf("%v", n)
+		} else if n, ok := pr["id"]; ok {
+			prNum = fmt.Sprintf("%v", n)
+		} else if n, ok := pr["number"]; ok {
+			prNum = fmt.Sprintf("%v", n)
+		} else if n, ok := pr["pull_request_id"]; ok {
+			prNum = fmt.Sprintf("%v", n)
+		}
+		if prNum == "" {
+			continue
+		}
+		bin := wf.ResolveCLIBinary()
+		cmd := exec.Command(bin, "pr", "+diff", "--id", prNum, "--owner", owner, "--repo", repo, "--format", "json")
+		out, err := cmd.Output()
+		if err != nil {
+			continue
+		}
+		var resp struct {
+			OK   bool `json:"ok"`
+			Data struct {
+				Files []struct {
+					Sections []struct {
+						Lines []struct {
+							Content string `json:"content"`
+						} `json:"lines"`
+					} `json:"sections"`
+				} `json:"files"`
+			} `json:"data"`
+		}
+		if err := json.Unmarshal(out, &resp); err != nil || !resp.OK || len(resp.Data.Files) == 0 {
+			continue
+		}
+		var sb strings.Builder
+		for _, f := range resp.Data.Files {
+			for _, sec := range f.Sections {
+				for _, line := range sec.Lines {
+					sb.WriteString(line.Content)
+					sb.WriteString("\n")
+				}
+			}
+		}
+		diffText := sb.String()
+		if diffText == "" {
+			continue
+		}
+		diffs[prNum] = diffText
+	}
+	if len(diffs) > 0 {
+		upstream["_pr_diffs"] = diffs
+	}
+}
+
+// prFingerprint returns an MD5 hash of key PR fields for change detection.
+func prFingerprint(pr map[string]interface{}) string {
+	var parts []string
+	if t := strFromMap(pr, "title", "name"); t != "" {
+		parts = append(parts, "title:"+t)
+	}
+	if b := strFromMap(pr, "body", "description"); b != "" {
+		parts = append(parts, "body:"+b)
+	}
+	if s := strFromMap(pr, "pull_request_status", "pull_request_staus", "status", "state"); s != "" {
+		parts = append(parts, "status:"+s)
+	}
+	h := md5.Sum([]byte(strings.Join(parts, "|")))
+	return hex.EncodeToString(h[:])
+}
+
+// prNumberFromMap extracts a PR number string from a PR data map.
+func prNumberFromMap(pr map[string]interface{}) string {
+	for _, k := range []string{"pull_request_number", "id", "number", "pull_request_id"} {
+		if v := pr[k]; v != nil {
+			s := fmt.Sprintf("%v", v)
+			if s != "" && s != "0" && s != "<nil>" {
+				return s
+			}
+		}
+	}
+	return ""
+}
+
+// strFromMap returns the first non-empty string value from the given keys.
+func strFromMap(m map[string]interface{}, keys ...string) string {
+	for _, k := range keys {
+		if v, ok := m[k].(string); ok && v != "" {
+			return v
+		}
+	}
+	return ""
+}
+
+// extractPRListFromUpstream extracts PR list from upstream data.
+func extractPRListFromUpstream(upstream map[string]interface{}, key string) []map[string]interface{} {
+	raw, ok := upstream[key]
+	if !ok {
+		return nil
+	}
+	if m, ok := raw.(map[string]interface{}); ok {
+		if data, ok := m["data"]; ok {
+			raw = data
+		}
+	}
+	if m, ok := raw.(map[string]interface{}); ok {
+		for _, listKey := range []string{"issues", "pull_requests"} {
+			if v, ok := m[listKey]; ok {
+				if arr, ok := v.([]interface{}); ok {
+					raw = arr
+					break
+				}
+			}
+		}
+	}
+	list, _ := raw.([]interface{})
+	var out []map[string]interface{}
+	for _, item := range list {
+		if m, ok := item.(map[string]interface{}); ok {
+			out = append(out, m)
+		}
+	}
+	return out
+}

shortcuts/workflow/state/state.go

@@ -0,0 +1,109 @@
+package state
+
+import (
+	"crypto/md5"
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"time"
+
+	"github.com/gitlink-org/gitlink-cli/internal/config"
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+// WorkflowState tracks persistent run state and change detection snapshots.
+type WorkflowState struct {
+	Workflow      string            `json:"workflow"`
+	LastRun       string            `json:"last_run"`
+	TotalRuns     int               `json:"total_runs"`
+	Snapshots     map[string]string `json:"snapshots"`
+	PhaseLastRun  map[string]string `json:"phase_last_run,omitempty"`
+	PhaseUpstream map[string]string `json:"phase_upstream,omitempty"`
+	ReviewedPRs   map[string]string `json:"reviewed_prs,omitempty"`
+}
+
+// LoadState reads the persisted workflow state from disk.
+func LoadState(name string) (*WorkflowState, error) {
+	path := statePath(name)
+	data, err := os.ReadFile(path)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return &WorkflowState{
+				Workflow:      name,
+				Snapshots:     make(map[string]string),
+				PhaseLastRun:  make(map[string]string),
+				PhaseUpstream: make(map[string]string),
+			}, nil
+		}
+		return nil, err
+	}
+	var s WorkflowState
+	if err := json.Unmarshal(data, &s); err != nil {
+		return nil, fmt.Errorf("parse state file %s: %w", path, err)
+	}
+	if s.Snapshots == nil {
+		s.Snapshots = make(map[string]string)
+	}
+	if s.PhaseLastRun == nil {
+		s.PhaseLastRun = make(map[string]string)
+	}
+	if s.PhaseUpstream == nil {
+		s.PhaseUpstream = make(map[string]string)
+	}
+	return &s, nil
+}
+
+// Save persists the workflow state to disk.
+func (s *WorkflowState) Save() error {
+	s.LastRun = time.Now().Format(time.RFC3339)
+	path := statePath(s.Workflow)
+	dir := filepath.Dir(path)
+	if err := os.MkdirAll(dir, 0700); err != nil {
+		return err
+	}
+	data, err := json.MarshalIndent(s, "", "  ")
+	if err != nil {
+		return err
+	}
+	return os.WriteFile(path, data, 0600)
+}
+
+// Diff compares current step results against stored snapshots.
+func (s *WorkflowState) Diff(results []wf.StepResult) []string {
+	changed := []string{}
+	for _, sr := range results {
+		if !sr.OK || sr.Data == nil {
+			continue
+		}
+		hash := HashData(sr.Data)
+		if prev, ok := s.Snapshots[sr.Step]; ok && prev != hash {
+			changed = append(changed, sr.Step)
+		}
+	}
+	return changed
+}
+
+// UpdateSnapshots stores hashes of current step results for future diff.
+func (s *WorkflowState) UpdateSnapshots(results []wf.StepResult) {
+	for _, sr := range results {
+		if !sr.OK || sr.Data == nil {
+			continue
+		}
+		s.Snapshots[sr.Step] = HashData(sr.Data)
+	}
+}
+
+// HashData computes an MD5 hash of the JSON-encoded data.
+func HashData(data interface{}) string {
+	b, err := json.Marshal(data)
+	if err != nil {
+		return ""
+	}
+	return fmt.Sprintf("%x", md5.Sum(b))
+}
+
+// statePath returns the file path for a workflow's state file.
+func statePath(name string) string {
+	return filepath.Join(config.ConfigDir(), fmt.Sprintf("workflow-%s-state.json", name))
+}

shortcuts/workflow/state/state_test.go

@@ -0,0 +1,80 @@
+package state
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	wf "github.com/gitlink-org/gitlink-cli/shortcuts/workflow"
+)
+
+func TestStateSaveLoad(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("GITLINK_CONFIG_DIR", dir)
+
+	s := &WorkflowState{
+		Workflow:  "test-wf",
+		TotalRuns: 5,
+		Snapshots: map[string]string{"step1": "abc123"},
+	}
+	if err := s.Save(); err != nil {
+		t.Fatalf("Save failed: %v", err)
+	}
+
+	loaded, err := LoadState("test-wf")
+	if err != nil {
+		t.Fatalf("LoadState failed: %v", err)
+	}
+	if loaded.TotalRuns != 5 {
+		t.Fatalf("TotalRuns = %d, want 5", loaded.TotalRuns)
+	}
+	if loaded.Snapshots["step1"] != "abc123" {
+		t.Fatalf("Snapshots[step1] = %q, want abc123", loaded.Snapshots["step1"])
+	}
+
+	os.Remove(filepath.Join(dir, "workflow-test-wf-state.json"))
+}
+
+func TestStateDiff(t *testing.T) {
+	s := &WorkflowState{
+		Workflow:  "test-diff",
+		Snapshots: map[string]string{"step1": "oldhash"},
+	}
+
+	results := []wf.StepResult{
+		{Step: "step1", OK: true, Data: "changed data"},
+		{Step: "step2", OK: true, Data: "new step"},
+		{Step: "step3", OK: false, Data: "ignored"},
+	}
+
+	changed := s.Diff(results)
+	if len(changed) != 1 {
+		t.Fatalf("Diff: expected 1 changed step, got %d", len(changed))
+	}
+	if changed[0] != "step1" {
+		t.Fatalf("Diff: expected 'step1' to change, got %q", changed[0])
+	}
+	s.UpdateSnapshots(results)
+	if _, ok := s.Snapshots["step2"]; !ok {
+		t.Fatal("step2 should be added to snapshots after UpdateSnapshots")
+	}
+	if _, ok := s.Snapshots["step3"]; ok {
+		t.Fatal("step3 (failed) should NOT be added to snapshots")
+	}
+}
+
+func TestLoadStateNotExist(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("GITLINK_CONFIG_DIR", dir)
+
+	s, err := LoadState("nonexistent")
+	if err != nil {
+		t.Fatalf("LoadState should not error for missing file: %v", err)
+	}
+	if s.Workflow != "nonexistent" {
+		t.Fatalf("Workflow = %q, want nonexistent", s.Workflow)
+	}
+	if s.Snapshots == nil {
+		t.Fatal("Snapshots should be initialized as empty map")
+	}
+}

shortcuts/workflow/types.go

@@ -0,0 +1,120 @@
+package workflow
+
+import (
+	"fmt"
+	"net/url"
+)
+
+// AIMode controls whether AI is used for skill steps.
+type AIMode string
+
+const (
+	AIModeAuto AIMode = "auto"
+	AIModeAI   AIMode = "ai"
+	AIModeNoAI AIMode = "no-ai"
+)
+
+// RuleEngineFunc is the signature for a deterministic rule engine.
+type RuleEngineFunc func(upstream map[string]interface{}, stepName string) (*AIResponse, error)
+
+// RuleEngines is a registry of skill-target → rule-engine mappings.
+var RuleEngines = map[string]RuleEngineFunc{}
+
+// RegisterRuleEngine registers a rule engine function for a given skill target.
+func RegisterRuleEngine(target string, fn RuleEngineFunc) {
+	RuleEngines[target] = fn
+}
+
+// ErrNoRuleEngine is returned when no rule engine is registered for a target.
+func ErrNoRuleEngine(target string) error {
+	return fmt.Errorf("no rule engine registered for skill target %q", target)
+}
+
+// StepType classifies what mechanism executes a step.
+type StepType string
+
+const (
+	StepTypeSkill   StepType = "skill"
+	StepTypeCommand StepType = "command"
+	StepTypeAPI     StepType = "api"
+)
+
+// RunWhen controls how often a step executes.
+type RunWhen string
+
+const (
+	RunAlways   RunWhen = "always"
+	RunWeekly   RunWhen = "weekly"
+	RunOnChange RunWhen = "on_change"
+)
+
+// StepDef defines a single step in a workflow.
+type StepDef struct {
+	Type      StepType   `json:"type"`
+	Name      string     `json:"name"`
+	Purpose   string     `json:"purpose"`
+	Target    string     `json:"target"`
+	DependsOn []string   `json:"depends_on,omitempty"`
+	RunWhen   RunWhen    `json:"run_when,omitempty"`
+	Method    string     `json:"method,omitempty"`
+	Query     url.Values `json:"-"`
+}
+
+// TriggerDef configures when a workflow runs.
+type TriggerDef struct {
+	Type     string `json:"type"`
+	On       string `json:"on"`
+	Interval string `json:"interval,omitempty"`
+}
+
+// WorkflowDef is a named, ordered sequence of steps with a trigger.
+type WorkflowDef struct {
+	Name        string     `json:"name"`
+	Category    string     `json:"category"`
+	Description string     `json:"description"`
+	Trigger     TriggerDef `json:"trigger"`
+	Steps       []StepDef  `json:"steps"`
+}
+
+// AIAction is a write instruction returned by an AI skill step.
+type AIAction struct {
+	Type    string                 `json:"type"`
+	Method  string                 `json:"method,omitempty"`
+	Path    string                 `json:"path,omitempty"`
+	Body    map[string]interface{} `json:"body,omitempty"`
+	Module  string                 `json:"module,omitempty"`
+	Command string                 `json:"command,omitempty"`
+	Args    map[string]string      `json:"args,omitempty"`
+}
+
+// AIRequest bundles the data needed for an AI skill step call.
+type AIRequest struct {
+	SystemPrompt string
+	UserData     string
+}
+
+// AIResponse is the parsed structured output from an AI skill step.
+type AIResponse struct {
+	Analysis interface{} `json:"analysis"`
+	Actions  []AIAction  `json:"actions"`
+}
+
+// WorkflowResult holds the outcome of a full workflow run.
+type WorkflowResult struct {
+	Workflow string       `json:"workflow"`
+	Owner    string       `json:"owner"`
+	Repo     string       `json:"repo"`
+	Steps    []StepResult `json:"steps"`
+}
+
+// StepResult holds the outcome of executing one step.
+type StepResult struct {
+	Step       string      `json:"step"`
+	Purpose    string      `json:"purpose"`
+	Type       StepType    `json:"type"`
+	OK         bool        `json:"ok"`
+	Data       interface{} `json:"data,omitempty"`
+	Error      string      `json:"error,omitempty"`
+	Skipped    bool        `json:"skipped,omitempty"`
+	SkipReason string      `json:"skip_reason,omitempty"`
+}

shortcuts/workflow/workflow_test.go

@@ -0,0 +1,259 @@
+package workflow
+
+import (
+	"testing"
+)
+
+func init() {
+	// Register all workflows for tests (defs init doesn't run in test context
+	// because importing defs would create an import cycle).
+	Register(&WorkflowDef{
+		Name:        "community-ops",
+		Category:    "运营",
+		Description: "社区运营自动化",
+		Trigger:     TriggerDef{Type: "poll", On: "issue.created", Interval: "5m"},
+		Steps: []StepDef{
+			{Type: StepTypeCommand, Name: "open-issues", Purpose: "获取开放 Issue", Target: "issue +list --state open --limit 50"},
+			{Type: StepTypeCommand, Name: "labels", Purpose: "获取标签库", Target: "label +list"},
+			{Type: StepTypeCommand, Name: "members", Purpose: "获取成员列表", Target: "member +list"},
+			{Type: StepTypeSkill, Name: "triage", Purpose: "AI 分析并执行分拣", Target: "gitlink-triage", DependsOn: []string{"open-issues", "labels", "members"}, RunWhen: RunAlways},
+			{Type: StepTypeCommand, Name: "repo-info", Purpose: "获取项目基础信息", Target: "repo +info"},
+			{Type: StepTypeCommand, Name: "merged-prs", Purpose: "获取已合并 PR", Target: "pr +list --state merged --limit 50"},
+			{Type: StepTypeCommand, Name: "commits", Purpose: "获取提交历史", Target: "commit +list --limit 50"},
+			{Type: StepTypeSkill, Name: "health-report", Purpose: "AI 生成周报", Target: "gitlink-health", DependsOn: []string{"repo-info", "merged-prs", "commits", "open-issues"}, RunWhen: RunWeekly},
+			{Type: StepTypeCommand, Name: "releases", Purpose: "获取版本发布记录", Target: "release +list"},
+			{Type: StepTypeSkill, Name: "changelog", Purpose: "生成 Release Notes", Target: "gitlink-changelog", DependsOn: []string{"commits", "releases", "merged-prs"}, RunWhen: RunOnChange},
+		},
+	})
+	Register(&WorkflowDef{
+		Name:        "code-quality",
+		Category:    "质量",
+		Description: "代码质量看门人",
+		Trigger:     TriggerDef{Type: "poll", On: "pr.opened", Interval: "5m"},
+		Steps: []StepDef{
+			{Type: StepTypeCommand, Name: "open-prs", Purpose: "获取开放 PR 列表", Target: "pr +list --state open --limit 20"},
+			{Type: StepTypeCommand, Name: "ci-builds", Purpose: "获取 CI 构建状态", Target: "ci +builds --limit 10"},
+			{Type: StepTypeCommand, Name: "repo-info", Purpose: "获取仓库保护规则配置", Target: "repo +info"},
+			{Type: StepTypeCommand, Name: "commits", Purpose: "获取最近提交", Target: "commit +list --limit 30"},
+			{Type: StepTypeCommand, Name: "branches", Purpose: "获取分支列表", Target: "branch +list"},
+			{Type: StepTypeSkill, Name: "review", Purpose: "AI 审查 PR 代码质量", Target: "gitlink-review", DependsOn: []string{"open-prs"}},
+			{Type: StepTypeSkill, Name: "ci-diagnosis", Purpose: "AI 诊断 CI 构建失败", Target: "gitlink-ci", DependsOn: []string{"ci-builds", "commits"}},
+		},
+	})
+	Register(&WorkflowDef{
+		Name:        "project-init",
+		Category:    "初始化",
+		Description: "项目一键初始化",
+		Trigger:     TriggerDef{Type: "manual", On: "manual"},
+		Steps: []StepDef{
+			{Type: StepTypeSkill, Name: "init-scaffold", Purpose: "创建仓库并初始化脚手架", Target: "gitlink-init-scaffold"},
+			{Type: StepTypeCommand, Name: "repo-info", Purpose: "确认仓库已创建", Target: "repo +info"},
+			{Type: StepTypeCommand, Name: "existing-files", Purpose: "验证文件", Target: "file +list"},
+			{Type: StepTypeCommand, Name: "labels", Purpose: "验证标签库", Target: "label +list"},
+			{Type: StepTypeSkill, Name: "license-check", Purpose: "检查许可证合规", Target: "gitlink-license", DependsOn: []string{"existing-files"}},
+			{Type: StepTypeCommand, Name: "milestones", Purpose: "验证里程碑", Target: "milestone +list"},
+			{Type: StepTypeCommand, Name: "existing-issues", Purpose: "验证初始 Issue", Target: "issue +list --state all --limit 10"},
+			{Type: StepTypeCommand, Name: "branches", Purpose: "检查分支结构", Target: "branch +list"},
+			{Type: StepTypeSkill, Name: "repo-audit", Purpose: "综合审计仓库健康度", Target: "gitlink-repo", DependsOn: []string{"repo-info", "existing-files", "labels", "milestones", "branches"}},
+		},
+	})
+	Register(&WorkflowDef{
+		Name:        "multi-repo",
+		Category:    "协同",
+		Description: "多仓库协同",
+		Trigger:     TriggerDef{Type: "cron", On: "0 9 * * 1", Interval: "24h"},
+		Steps: []StepDef{
+			{Type: StepTypeCommand, Name: "multi-repo-snapshot", Purpose: "采集多个仓库状态", Target: "workflow-internal:multi-repo-snapshot"},
+			{Type: StepTypeSkill, Name: "multi-repo-coordination", Purpose: "生成统一报告", Target: "gitlink-multi-repo", DependsOn: []string{"multi-repo-snapshot"}},
+		},
+	})
+	Register(&WorkflowDef{
+		Name:        "contributor-growth",
+		Category:    "成长",
+		Description: "贡献者成长体系",
+		Trigger:     TriggerDef{Type: "cron", On: "0 9 * * 1", Interval: "24h"},
+		Steps: []StepDef{
+			{Type: StepTypeCommand, Name: "commits", Purpose: "提交历史", Target: "commit +list --limit 100"},
+			{Type: StepTypeCommand, Name: "open-issues", Purpose: "开放 Issue", Target: "issue +list --state open --limit 50"},
+			{Type: StepTypeCommand, Name: "closed-issues", Purpose: "已关闭 Issue", Target: "issue +list --state closed --limit 50"},
+			{Type: StepTypeCommand, Name: "merged-prs", Purpose: "已合并 PR", Target: "pr +list --state merged --limit 50"},
+			{Type: StepTypeCommand, Name: "members", Purpose: "项目成员列表", Target: "member +list"},
+			{Type: StepTypeCommand, Name: "repo-info", Purpose: "项目基础数据", Target: "repo +info"},
+			{Type: StepTypeSkill, Name: "contributor-ranking", Purpose: "AI 分析贡献者排行", Target: "gitlink-contributor-ranking", DependsOn: []string{"commits", "open-issues", "closed-issues", "merged-prs", "members"}},
+		},
+	})
+}
+
+func TestRegistry(t *testing.T) {
+	if len(registry) != 5 {
+		t.Fatalf("expected 5 workflows, got %d", len(registry))
+	}
+
+	for _, name := range []string{"community-ops", "code-quality", "project-init", "multi-repo", "contributor-growth"} {
+		wf := Get(name)
+		if wf == nil {
+			t.Fatalf("workflow %q not found", name)
+		}
+		if len(wf.Steps) == 0 {
+			t.Fatalf("workflow %q has no steps", name)
+		}
+		if wf.Trigger.Type == "" {
+			t.Fatalf("workflow %q has no trigger.type", name)
+		}
+	}
+
+	all := All()
+	if len(all) != 5 {
+		t.Fatalf("All() returned %d workflows, expected 5", len(all))
+	}
+}
+
+func TestGetNonexistent(t *testing.T) {
+	if Get("nonexistent") != nil {
+		t.Fatal("expected nil for nonexistent workflow")
+	}
+}
+
+func TestResolvePath(t *testing.T) {
+	cases := []struct {
+		template, owner, repo, expected string
+	}{
+		{"{v1}/issues", "chroe", "gitlink-cli", "/v1/chroe/gitlink-cli/issues"},
+		{"{base}/pulls", "chroe", "gitlink-cli", "/chroe/gitlink-cli/pulls"},
+		{"{base}", "org", "proj", "/org/proj"},
+		{"{v1}/issues?state=open", "x", "y", "/v1/x/y/issues?state=open"},
+	}
+	for _, tc := range cases {
+		got := ResolvePath(tc.template, tc.owner, tc.repo)
+		if got != tc.expected {
+			t.Fatalf("ResolvePath(%q, %s, %s) = %q, want %q", tc.template, tc.owner, tc.repo, got, tc.expected)
+		}
+	}
+}
+
+func TestTriggers(t *testing.T) {
+	expected := map[string]struct {
+		on  string
+		typ string
+	}{
+		"community-ops":      {"issue.created", "poll"},
+		"code-quality":       {"pr.opened", "poll"},
+		"project-init":       {"manual", "manual"},
+		"multi-repo":         {"0 9 * * 1", "cron"},
+		"contributor-growth": {"0 9 * * 1", "cron"},
+	}
+	for name, want := range expected {
+		wf := Get(name)
+		if wf.Trigger.On != want.on {
+			t.Fatalf("%s: trigger.on = %q, want %q", name, wf.Trigger.On, want.on)
+		}
+		if wf.Trigger.Type != want.typ {
+			t.Fatalf("%s: trigger.type = %q, want %q", name, wf.Trigger.Type, want.typ)
+		}
+	}
+}
+
+func TestStepTypes(t *testing.T) {
+	wf := Get("community-ops")
+	if wf == nil {
+		t.Fatal("community-ops not found")
+	}
+
+	typeCounts := map[StepType]int{}
+	for _, s := range wf.Steps {
+		typeCounts[s.Type]++
+	}
+	if typeCounts[StepTypeCommand] < 1 {
+		t.Fatal("community-ops should have at least one command step")
+	}
+	if typeCounts[StepTypeSkill] < 1 {
+		t.Fatal("community-ops should have at least one skill step")
+	}
+}
+
+func TestSkillStepDependsOn(t *testing.T) {
+	wf := Get("community-ops")
+	if wf == nil {
+		t.Fatal("community-ops not found")
+	}
+
+	var triage *StepDef
+	for i := range wf.Steps {
+		if wf.Steps[i].Name == "triage" {
+			triage = &wf.Steps[i]
+			break
+		}
+	}
+	if triage == nil {
+		t.Fatal("triage step not found")
+	}
+	if len(triage.DependsOn) != 3 {
+		t.Fatalf("triage step should have 3 dependencies, got %d: %v", len(triage.DependsOn), triage.DependsOn)
+	}
+	expectedDeps := map[string]bool{"open-issues": false, "labels": false, "members": false}
+	for _, dep := range triage.DependsOn {
+		if _, ok := expectedDeps[dep]; !ok {
+			t.Fatalf("unexpected dependency: %s", dep)
+		}
+		expectedDeps[dep] = true
+	}
+}
+
+func TestMultiRepoWorkflowShape(t *testing.T) {
+	wf := Get("multi-repo")
+	if wf == nil {
+		t.Fatal("multi-repo not found")
+	}
+	if len(wf.Steps) != 2 {
+		t.Fatalf("multi-repo should have 2 steps, got %d", len(wf.Steps))
+	}
+	if wf.Steps[0].Name != "multi-repo-snapshot" || wf.Steps[0].Target != "workflow-internal:multi-repo-snapshot" {
+		t.Fatalf("unexpected snapshot step: %+v", wf.Steps[0])
+	}
+	if wf.Steps[1].Target != "gitlink-multi-repo" {
+		t.Fatalf("multi-repo skill target = %q, want gitlink-multi-repo", wf.Steps[1].Target)
+	}
+}
+
+func TestParseCommandTarget(t *testing.T) {
+	cases := []struct {
+		input    string
+		expected []string
+	}{
+		{"issue +list --state open", []string{"issue", "+list", "--state", "open"}},
+		{"repo +info", []string{"repo", "+info"}},
+		{"pr +list --state merged --limit 50", []string{"pr", "+list", "--state", "merged", "--limit", "50"}},
+		{"issue +list --state open --limit 50", []string{"issue", "+list", "--state", "open", "--limit", "50"}},
+	}
+	for _, tc := range cases {
+		got := ParseCommandTarget(tc.input)
+		if len(got) != len(tc.expected) {
+			t.Fatalf("ParseCommandTarget(%q): len=%d, want len=%d (got=%v)", tc.input, len(got), len(tc.expected), got)
+		}
+		for i := range got {
+			if got[i] != tc.expected[i] {
+				t.Fatalf("ParseCommandTarget(%q)[%d] = %q, want %q", tc.input, i, got[i], tc.expected[i])
+			}
+		}
+	}
+}
+
+func TestCodeQualityHasReviewStep(t *testing.T) {
+	wf := Get("code-quality")
+	if wf == nil {
+		t.Fatal("code-quality not found")
+	}
+	if len(wf.Steps) < 7 {
+		t.Fatalf("code-quality should have at least 7 steps (including review), got %d", len(wf.Steps))
+	}
+	found := false
+	for _, s := range wf.Steps {
+		if s.Target == "gitlink-review" {
+			found = true
+			break
+		}
+	}
+	if !found {
+		t.Fatal("code-quality missing gitlink-review skill step")
+	}
+}

showcase/Dockerfile

@@ -1,3 +1,4 @@
+# syntax=docker/dockerfile:1
 # Stage 1: Build
 FROM golang:1.26-alpine AS builder
 
@@ -9,17 +10,20 @@ RUN go mod download
 
 COPY . .
 
-# Build gitlink-cli binary
-RUN CGO_ENABLED=0 GOOS=linux go build -o gitlink-cli .
-
-# Build showcase server binary
-RUN CGO_ENABLED=0 GOOS=linux go build -o showcase-server ./showcase/
+# 合并为一个 RUN：showcase-server 复用 gitlink-cli 已编译的依赖（同一 RUN 层内 Go 编译缓存共享）。
+# --mount=type=cache：BuildKit 跨构建保留 Go 编译缓存（/root/.cache/go-build），
+# 之后只重编改动的包，不再全量重编（需 docker build 时 DOCKER_BUILDKIT=1）。
+RUN --mount=type=cache,target=/root/.cache/go-build \
+    CGO_ENABLED=0 GOOS=linux go build -buildvcs=false -o gitlink-cli . && \
+    CGO_ENABLED=0 GOOS=linux go build -buildvcs=false -o showcase-server ./showcase/
 
 # Stage 2: Runtime
 FROM alpine:latest
 
 RUN apk add --no-cache ca-certificates git
 
+RUN mkdir -p /root/.config/gitlink-cli
+
 WORKDIR /app
 
 COPY --from=builder /build/gitlink-cli .

showcase/main.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+	"bytes"
 	_ "embed"
 	"encoding/json"
 	"fmt"
@@ -31,6 +32,7 @@ func main() {
 
 	http.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Content-Type", "text/html; charset=utf-8")
+		w.Header().Set("Cache-Control", "no-cache, no-store, must-revalidate")
 		w.Write(indexHTML)
 	})
 
@@ -41,7 +43,9 @@ func main() {
 		command := r.URL.Query().Get("command")
 		owner := r.URL.Query().Get("owner")
 		repo := r.URL.Query().Get("repo")
+		format := r.URL.Query().Get("format")
 		extraArgs := r.URL.Query().Get("args")
+		descParam := r.URL.Query().Get("desc")
 
 		if module == "" || command == "" {
 			json.NewEncoder(w).Encode(RunResult{Error: "missing module or command"})
@@ -58,31 +62,67 @@ func main() {
 			}
 		}
 
-		args := []string{module, "+" + command, "--owner", owner, "--repo", repo, "--format", "json"}
+		args := []string{module, "+" + command, "--owner", owner, "--format", format}
+		if !strings.Contains(extraArgs, "project-init") && !strings.Contains(extraArgs, "multi-repo") {
+			args = append(args, "--repo", repo)
+		}
 		if extraArgs != "" {
 			args = append(args, parseShellArgs(extraArgs)...)
 		}
+		// If desc is passed as a dedicated query parameter, ensure it is
+		// passed intact (bypasses the space-splitting in parseShellArgs).
+		if descParam != "" {
+			// Remove any --desc from parseShellArgs output (old JS path).
+			filtered := make([]string, 0, len(args))
+			skipNext := false
+			for i, a := range args {
+				if skipNext {
+					skipNext = false
+					continue
+				}
+				if a == "--desc" || a == "--description" {
+					skipNext = true
+					continue
+				}
+				if strings.HasPrefix(a, "--desc=") || strings.HasPrefix(a, "--description=") {
+					continue
+				}
+				_ = i
+				filtered = append(filtered, a)
+			}
+			args = filtered
+			args = append(args, "--desc", descParam)
+		}
 
 		cmdStr := "gitlink-cli " + strings.Join(args, " ")
 		log.Printf("Running: %s", cmdStr)
 
 		cmd := exec.Command(cliBin, args...)
-		output, err := cmd.CombinedOutput()
+		var stdout, stderr bytes.Buffer
+		cmd.Stdout = &stdout
+		cmd.Stderr = &stderr
+		runErr := cmd.Run()
 
 		result := RunResult{
 			Command: cmdStr,
 		}
 
-		if err != nil {
-			result.Error = strings.TrimSpace(string(output))
+		if runErr != nil {
+			result.Error = strings.TrimSpace(stderr.String())
+			if result.Error == "" {
+				result.Error = strings.TrimSpace(stdout.String())
+			}
+			if result.Error == "" {
+				result.Error = runErr.Error()
+			}
 			result.Output = nil
 		} else {
 			result.OK = true
 			var parsed interface{}
-			if json.Unmarshal(output, &parsed) == nil {
+			if json.Unmarshal(stdout.Bytes(), &parsed) == nil {
 				result.Output = parsed
 			} else {
-				result.Output = strings.TrimSpace(string(output))
+				result.Output = strings.TrimSpace(stdout.String())
 			}
 		}
 
@@ -93,8 +133,6 @@ func main() {
 	log.Fatal(http.ListenAndServe(":"+port, nil))
 }
 
-// parseShellArgs splits a shell-style argument string, respecting quoted values.
-// e.g. `--content "Hello Wiki!" --message "create page"` -> ["--content", "Hello Wiki!", "--message", "create page"]
 func parseShellArgs(s string) []string {
 	var args []string
 	var current strings.Builder

skills/README.md

@@ -108,8 +108,24 @@ skills/
 │       └── ci-workflow.md             # CI 工作流
 ├── gitlink-pm/                        # 项目管理
 │   └── SKILL.md                       # PM 操作指南
-└── gitlink-workflow/                  # AI 自动化工作流
-    └── SKILL.md                       # 工作流模板（Issue 分类、PR Review、Release Notes）
+├── gitlink-workflow/                  # AI 自动化工作流
+│   └── SKILL.md                       # 工作流模板（Issue 分类、PR Review、Release Notes）
+├── gitlink-health/                    # 项目健康度报告
+│   ├── SKILL.md                       # 健康度报告操作指南
+│   ├── REFERENCE.md                   # 指标体系与评分算法
+│   └── examples/
+│       ├── health-report.md           # 完整健康度报告示例
+│       └── weekly-report.md           # 周报模式示例
+├── gitlink-changelog/                 # Release Notes 自动生成
+│   ├── SKILL.md                       # Changelog 操作指南
+│   ├── REFERENCE.md                   # Commit 分类规则与字段映射
+│   └── examples/
+│       └── generate-release-notes.md  # Release Notes 生成示例
+└── gitlink-triage/                    # Issue 智能分拣与新人引导
+    ├── SKILL.md                       # 分拣操作指南
+    ├── REFERENCE.md                   # 标签体系与分配算法
+    └── examples/
+        └── batch-triage.md            # 批量分拣示例
 ```
 
 ---
@@ -138,6 +154,17 @@ skills/
 | **gitlink-pm** | 项目管理 | 通过 Raw API 访问 |
 | **gitlink-workflow** | AI 工作流 | Issue 分类、PR Review、Release Notes |
 
+### 🆕 智能 Skills（v1.2 新增）
+
+| Skill | 说明 | 场景 |
+|-------|------|------|
+| **gitlink-health** | 项目健康度报告 | Issue 响应时间、PR 合并效率、贡献者活跃度统计 |
+| **gitlink-changelog** | Release Notes 自动生成 | 从 commit/PR/Issue 历史自动生成版本说明 |
+| **gitlink-triage** | Issue 智能分拣 + 新人引导 | 自动分类、打标签、分配责任人、good-first-issue 引导 |
+| **gitlink-review** | 智能代码审查 | 分析 PR diff，多视角评审 + 对抗式自检，结构化 Review 意见自动评论 |
+| **gitlink-research-fair** | 科研软件 X 光 | fair.py 抽论文/数据/复现/引用画像，四维裁决 + 真科研图谱 + 特有关键发现报告，可选处方 PR |
+| **gitlink-spark** | 文献-代码语义缺口挖掘机 | arXiv×GitLink×GitHub 三源挖"理论无实现/需求无解答"缺口，出机会报告，一键 fork+issue 起跑 |
+
 ---
 
 ## 🎯 使用场景
@@ -304,6 +331,23 @@ AI 代理可以：
 
 ---
 
+## 🤖 在 Claude Code 中通过 Skill 工具调用
+
+默认情况下，这些 Skill 以**文档驱动**方式使用：AI 代理读取 `SKILL.md` 后执行 `gitlink-cli` 命令（见上文流程图）。
+
+若希望像标准 Skill 一样，由 Claude Code 的 **Skill 工具**直接加载并调用（例如 `Skill gitlink-issue`），需要把 `skills/gitlink-*` 链接到 Claude Code 扫描的 `~/.claude/skills/` 目录。
+
+仓库提供一键脚本，自动检测平台并创建链接（**Windows 用目录联接 junction，无需管理员权限；macOS/Linux 用 symlink**），均指向仓库内的 `skills/`，更新仓库后内容自动同步：
+
+```bash
+# 在仓库根目录执行（Windows 需 Git Bash 或 WSL）
+bash scripts/setup-skills.sh
+```
+
+执行后即可在 Claude Code 中通过 Skill 工具调用任意 `gitlink-*` Skill。脚本可重复执行，会安全更新已有链接——移除时只删 junction/symlink 本身，**不会影响 `skills/` 下的源文件**。
+
+---
+
 ## 📊 测试状态
 
 ✅ **生产就绪** (8.5/10)

skills/gitlink-branch/REFERENCE.md

@@ -0,0 +1,109 @@
+# gitlink-branch 参考手册
+
+> 本文档定义分支管理命令的参数说明、API 字段映射和注意事项。
+
+---
+
+## 一、命令参考
+
+### branch +list
+
+列出仓库所有分支。
+
+```bash
+gitlink-cli branch +list --owner <owner> --repo <repo> --format json
+```
+
+返回字段：
+
+| 字段 | 说明 |
+|------|------|
+| `name` | 分支名称 |
+| `protected` | 是否受保护（true/false） |
+
+实测返回（chroe/gitlink-cli）：
+
+```
+master, test/pr-workflow-demo, pr/showcase-dashboard, pr/shortcuts-modules,
+fix/showcase-member-v2, fix/showcase-member-org, fix/batch-invite-userid,
+feat/task-a-shortcuts, fix/windows-npm-install
+共 9 个分支
+```
+
+### branch +create
+
+从指定分支或 commit 创建新分支。
+
+```bash
+gitlink-cli branch +create --owner <owner> --repo <repo> \
+  --name <新分支名> --from <源分支> --format json
+```
+
+| 参数 | 必填 | 说明 |
+|------|------|------|
+| `--name, -n` | 是 | 新分支名称 |
+| `--from, -f` | 否 | 源分支或 commit（默认 master） |
+
+API：
+
+```
+POST /v1/{owner}/{repo}/branches.json
+Body: { "branch_name": name, "ref": from }
+```
+
+### branch +delete
+
+删除指定分支。
+
+```bash
+gitlink-cli branch +delete --owner <owner> --repo <repo> \
+  --name <分支名> --format json
+```
+
+| 参数 | 必填 | 说明 |
+|------|------|------|
+| `--name, -n` | 是 | 要删除的分支名称 |
+
+> ⚠️ **写入操作** — 删除分支不可恢复，执行前确认用户意图。不能删除默认分支。
+
+### branch +protect
+
+设置分支保护。
+
+```bash
+gitlink-cli branch +protect --owner <owner> --repo <repo> \
+  --name <分支名> --format json
+```
+
+| 参数 | 必填 | 说明 |
+|------|------|------|
+| `--name, -n` | 是 | 要保护的分支名称 |
+
+> ⚠️ 受保护的分支不允许直接 push 和强制推送。
+
+---
+
+## 二、分支命名约定
+
+| 前缀 | 用途 | 示例 |
+|------|------|------|
+| `feat/` | 新功能 | `feat/task-a-shortcuts` |
+| `fix/` | Bug 修复 | `fix/showcase-member-v2` |
+| `pr/` | PR 相关 | `pr/showcase-dashboard` |
+| `test/` | 测试 | `test/pr-workflow-demo` |
+
+---
+
+## 三、常见问题
+
+### Q: branch +list 返回的分支数量很多怎么办？
+
+A: 正常现象，开发过程中会积累功能分支。定期清理已合并的分支。
+
+### Q: 能删除 master 分支吗？
+
+A: 不能。GitLink 不允许删除默认分支。
+
+### Q: branch +protect 和仓库设置里的保护有什么区别？
+
+A: 效果相同，CLI 命令是仓库设置的快捷方式。

skills/gitlink-branch/examples/branch-workflow.md

@@ -1,70 +1,44 @@
-# 分支管理完整工作流
+# 示例：分支管理（真实数据）
 
-本示例演示一个完整的分支管理流程：查看分支 → 创建分支 → 开发推送 → 保护分支 → 清理旧分支。
+> 本示例基于 `chroe/gitlink-cli` 项目于 2026-06-09 在 Claude Code 中实际执行。
 
-## 场景描述
+---
 
-你正在维护一个 GitLink 仓库，需要开发一个新功能。开发完成后需要保护主分支，并清理已完成的功能分支。
-
-## 步骤
-
-### 1. 查看当前所有分支
+## 执行命令序列
 
 ```bash
-gitlink-cli branch +list
+# Step 1: 查看所有分支
+gitlink-cli branch +list --owner chroe --repo gitlink-cli --format json
+# 结果: 9 个分支
+#   master                          protected=False
+#   test/pr-workflow-demo           protected=False
+#   pr/showcase-dashboard           protected=False
+#   pr/shortcuts-modules            protected=False
+#   fix/showcase-member-v2          protected=False
+#   fix/showcase-member-org         protected=False
+#   fix/batch-invite-userid         protected=False
+#   feat/task-a-shortcuts           protected=False
+#   fix/windows-npm-install         protected=False
+
+# Step 2: 保护 master 分支
+gitlink-cli branch +protect --owner chroe --repo gitlink-cli --name master --format json
+# 结果: ok=True
+#   branch_name=master, created_at=2026-06-13 10:47
 ```
 
-输出示例：
-```
-Branch            Protected
-master            Yes
-develop           No
-feature/old-login No
+## 完整工作流
+
 ```
+用户："帮我看看仓库有哪些分支，把 master 保护一下"
 
-### 2. 创建新功能分支
-
-```bash
-gitlink-cli branch +create --name feature/new-auth --from develop
-```
-
-输出确认分支创建成功。
-
-### 3. 本地开发并推送
-
-```bash
-git checkout -b feature/new-auth
-# ... 编写代码 ...
-git add .
-git commit -m "feat: add new auth module"
-git push gitlink feature/new-auth
-```
-
-### 4. 创建 PR 并合并
-
-通过 `gitlink-cli pr +create` 创建 PR，审查后合并。合并后，功能分支 `feature/new-auth` 将自动删除（取决于仓库设置）。
-
-### 5. 保护主分支
-
-确保 `master` 分支有保护规则，防止误删或直接推送：
-
-```bash
-gitlink-cli branch +protect --name master
-```
-
-### 6. 清理旧分支（可选）
-
-开发完成后，删除不再需要的旧分支：
-
-```bash
-# 确认旧分支已合并或无保留价值后
-gitlink-cli branch +delete --name feature/old-login
+AI 执行流程：
+1. branch +list → 列出所有分支，发现 9 个分支，master 未保护
+2. branch +protect --name master → 保护 master 分支
+3. 报告结果：master 已保护，其他 8 个功能分支可以按需清理
 ```
 
 ## 注意事项
 
-- `branch +delete` 是 **Destructive Operation**，操作前请确认分支内容已合并或无保留价值
-- `branch +protect` 会对分支施加保护规则，影响后续的推送和合并流程
-- `branch +unprotect` 仅支持简单分支名（如 `main`），含 `/` 的路径需通过 Web 页面操作
-- 建议使用 `branch +list` 先查看分支状态，再执行写入/删除操作
-- 建议使用 `branch +list` 先查看分支状态，再执行写入/删除操作
+- `branch +delete` 是写入操作，执行前确认用户意图
+- 不能删除默认分支（master）
+- 建议先 `branch +list` 查看状态再执行写入操作

skills/gitlink-changelog/REFERENCE.md

@@ -0,0 +1,197 @@
+# gitlink-changelog 参考手册
+
+> 本文档定义 Release Notes 自动生成的分类规则、数据字段和 API 注意事项。
+
+---
+
+## 一、Commit 分类规则（详细版）
+
+### 主分类
+
+| 分类 | 图标 | 前缀关键词 | 优先级 |
+|------|------|-----------|--------|
+| 新功能 | ✨ | `feat`, `feature`, `add`, `新增`, `支持` | 高 |
+| Bug 修复 | 🐛 | `fix`, `bugfix`, `hotfix`, `修复`, `解决` | 高 |
+| 破坏性变更 | ⚠️ | 标题含 `!`（如 `feat!:`）或 body 含 `BREAKING CHANGE` | 最高 |
+| 改进优化 | 🔧 | `refactor`, `perf`, `improve`, `优化`, `增强`, `完善`, `enhance` | 中 |
+| 文档更新 | 📚 | `docs`, `doc`, `文档`, `README`, `注释` | 低 |
+| 测试 | 🧪 | `test`, `tests`, `测试` | 低 |
+| 构建/CI | 🏗️ | `build`, `ci`, `chore`, `构建`, `部署`, `Docker` | 低 |
+| 样式 | 💄 | `style`, `fmt`, `格式化` | 最低 |
+
+### 分类逻辑
+
+```
+1. 优先检测破坏性变更（!后缀 或 BREAKING CHANGE body）
+2. 检查前缀是否匹配已知分类
+3. 无前缀时，AI 根据 commit message 语义判断：
+   - 包含"新增"、"添加" → 新功能
+   - 包含"修复"、"解决"、"fix" → Bug 修复
+   - 包含"优化"、"改进"、"重构" → 改进优化
+   - 其他 → 归入"其他变更"分类
+```
+
+### 模块识别
+
+从 commit message 或 PR 标题中提取涉及的模块：
+
+| 模式 | 示例 | 提取模块 |
+|------|------|---------|
+| `feat(<module>):` | `feat(wiki): add create` | wiki |
+| `fix <module>:` | `fix batch operations:` | batch |
+| 中文括号 | `新增 wiki 模块` | wiki |
+| 文件路径 | `shortcuts/wiki/wiki.go` | wiki |
+
+---
+
+## 二、数据字段映射
+
+### commit +list 关键字段
+
+```json
+{
+  "sha": "abc123def456",
+  "message": "feat(wiki): add create and update commands",
+  "author": {
+    "login": "username",
+    "name": "用户名",
+    "image_url": "..."
+  },
+  "timestamp": "2026-06-01T10:00:00+08:00"
+}
+```
+
+| 字段 | 用途 |
+|------|------|
+| `message` | 分类依据，提取功能描述 |
+| `author.login` | 贡献者归属 |
+| `timestamp` | 判断是否属于当前版本区间 |
+| `sha` | 可作为变更追溯引用 |
+
+### pr +list / pr +view 关键字段
+
+```json
+{
+  "id": 42,
+  "title": "feat(wiki): add wiki management commands",
+  "body": "## 变更说明\n...",
+  "pull_request_status": 1,
+  "created_at": "2026-06-01T10:00:00+08:00",
+  "updated_at": "2026-06-02T15:00:00+08:00",
+  "merged_at": "2026-06-02T15:00:00+08:00",
+  "user": { "login": "username", "name": "用户名" },
+  "head": "feature/wiki",
+  "base": "master"
+}
+```
+
+| 字段 | 用途 |
+|------|------|
+| `title` | 分类依据 |
+| `body` | 提取详细变更说明 |
+| `merged_at` | 判断是否属于当前版本区间 |
+| `user.login` | 贡献者归属 |
+| `id` | PR 编号引用 |
+
+### issue +list 关键字段
+
+```json
+{
+  "id": 1,
+  "project_issues_index": 15,
+  "subject": "Wiki 中文路径乱码",
+  "description": "...",
+  "created_on": "2026-05-28T10:00:00+08:00",
+  "updated_on": "2026-06-01T15:00:00+08:00",
+  "status": { "id": 5, "name": "关闭" },
+  "author": { "login": "username" },
+  "tags": [{ "id": 1, "name": "bug" }]
+}
+```
+
+| 字段 | 用途 |
+|------|------|
+| `subject` | Issue 标题引用 |
+| `status.id === 5` | 已关闭 |
+| `updated_on` | 判断关闭时间是否在版本区间内 |
+| `tags` | 辅助分类（bug 标签 → Bug 修复） |
+| `project_issues_index` | Issue 编号引用（#15） |
+
+### release +list 关键字段
+
+```json
+{
+  "id": 3,
+  "tag_name": "v1.1.0",
+  "name": "v1.1.0",
+  "body": "## ✨ 新功能\n...",
+  "created_at": "2026-05-15T10:00:00+08:00"
+}
+```
+
+| 字段 | 用途 |
+|------|------|
+| `tag_name` | 版本号，确定版本基线 |
+| `created_at` | 版本发布时间，作为时间基线 |
+
+---
+
+## 三、版本区间计算
+
+### 方法 1：基于上一个 Release 的时间戳
+
+```
+版本区间开始 = release +list 中最新一条的 created_at
+版本区间结束 = 当前时间
+```
+
+### 方法 2：基于 Git Tag 对比（已废弃）
+
+> ⚠️ **此方法不可用。** compare API 实测返回 HTML 而非 JSON，无法使用。请改用方法 1 或方法 3。
+
+### 方法 3：无历史版本时
+
+```
+版本区间开始 = repo +info 的 created_on（项目创建时间）
+版本区间结束 = 当前时间
+```
+
+---
+
+## 四、Release Notes 发布注意事项
+
+1. **`release +create` 需要 `--tag` 参数**：即 Git tag 名称（如 `v1.2.0`）
+2. **`--body` 中的换行**：使用 `\n` 转义，或由 CLI 内部处理多行文本
+3. **`release +view` 使用 `version_id` 而非 tag_name**：创建后从 `release +list` 获取 `version_id`
+4. **发布前确认**：写入操作前必须确认用户意图
+5. **Tag 必须存在**：如果 tag 不存在于仓库，需要先通过 git 创建 tag 并推送
+
+---
+
+## 五、常见问题
+
+### Q: Commit message 不规范怎么办？
+
+> ⚠️ **注意**：compare API (`api GET /:owner/:repo/compare/...`) 实测返回 HTML 而非 JSON，不可使用。请改用 `commit +list` 按时间戳筛选版本区间内的提交。
+
+A: AI 根据 message 内容语义自动判断分类，无需严格遵循 Conventional Commits。但规范的 message 能显著提高分类准确度。
+
+### Q: 如何处理合并提交？
+
+A: 合并提交（merge commit）的 message 通常为 `Merge pull request #xx`，AI 应跳过这些提交，改为从对应的 PR 记录获取变更信息。
+
+### Q: 大量提交时如何处理？
+
+A: 使用分页参数逐步采集，AI 可先统计总量再决定是否需要完整采集。对于超过 100 条提交的版本，建议按模块汇总而非逐条列出。
+
+### Q: 如何生成英文 Release Notes？
+
+A: 在提示 AI 时指定"Generate Release Notes in English"。Skill 支持中英文输出，由 AI 根据用户语言偏好决定。
+
+### Q: PR `--state merged` 返回的数据准确吗？
+
+A: 注意 GitLink 的 `--state` 参数仅影响统计计数，返回列表可能包含所有状态的 PR。需要通过 `pull_request_status` 字段客户端过滤：`pull_request_status === 1` 才是真正已合并的 PR。
+
+### Q: 项目无历史 Release 时怎么确定版本区间？
+
+A: 使用 `repo +info` 获取项目创建时间（`created_on`），作为版本区间的起始时间。无需指定 base 版本。

skills/gitlink-changelog/SKILL.md

@@ -0,0 +1,209 @@
+---
+name: gitlink-changelog
+version: 1.0.0
+description: "Release Notes 自动生成：从 commit、PR、Issue 历史自动分析并生成结构化版本说明。当用户需要生成 Changelog、Release Notes、版本发布说明时触发。"
+metadata:
+  requires:
+    bins: ["gitlink-cli"]
+  cliHelp: "gitlink-cli --help"
+---
+
+# gitlink-changelog（Release Notes 自动生成）
+
+**CRITICAL — 开始前必须先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)，其中包含认证、权限处理和 API 注意事项。**
+**CRITICAL — 分类 commit 和生成 Release Notes 前，必须先阅读 [`REFERENCE.md`](REFERENCE.md)，其中包含完整的分类规则、优先级逻辑和 API 字段映射。**
+**CRITICAL — 本 Skill 的数据采集阶段仅读取数据；最终发布 Release 时为写入操作，发布前务必确认用户意图。**
+**CRITICAL — GitLink 操作只能用 `gitlink-cli`。禁止用 `gh`（GitHub CLI）操作 GitLink 资源。`gh` 仅适用于 GitHub 平台。**
+
+> **前置条件：** 先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md) 了解认证和全局参数。
+
+## 概述
+
+本 Skill 从 GitLink 项目的提交历史、PR 记录和 Issue 数据中自动提取版本变更信息，由 AI 分析分类后生成结构化的 Release Notes / Changelog。支持一键发布到 GitLink Release。
+
+## 数据采集命令
+
+### 第一步：获取版本基线
+
+```bash
+# 获取上一个 Release 信息（确定版本基线）
+gitlink-cli release +list --owner <owner> --repo <repo> --format json
+```
+
+### 第二步：获取提交历史
+
+```bash
+# 获取提交列表（AI 根据上一个 Release 的时间筛选新增提交）
+gitlink-cli commit +list --owner <owner> --repo <repo> --format json
+
+# 注意：提交可能需要翻页，用 --page 参数获取全部
+gitlink-cli commit +list --owner <owner> --repo <repo> --page 2 --format json
+```
+
+> ⚠️ **不要使用 compare API**。`api GET /:owner/:repo/compare/...` 实测会返回 HTML 而非 JSON，无法用于变更对比。请改用 `commit +list` 按时间戳筛选。
+
+### 第三步：获取 PR 记录
+
+```bash
+# 获取已合并的 PR 列表
+gitlink-cli pr +list --owner <owner> --repo <repo> --state merged --format json
+
+# 查看具体 PR 详情（获取关联 Issue 和变更内容）
+gitlink-cli pr +view --owner <owner> --repo <repo> --id <pr_id> --format json
+
+# 查看 PR 变更文件
+gitlink-cli pr +files --owner <owner> --repo <repo> --id <pr_id> --format json
+```
+
+### 第四步：获取已关闭 Issue
+
+```bash
+# 获取已关闭的 Issue
+gitlink-cli issue +list --owner <owner> --repo <repo> --state closed --format json
+```
+
+## Release Notes 输出格式
+
+AI 根据采集到的数据，按以下结构生成 Release Notes：
+
+```markdown
+# 🚀 <版本号> (<发布日期>)
+
+## 📝 版本摘要
+
+<一段话总结本次版本的核心变更>
+
+## ✨ 新功能 (Features)
+
+- **<模块>**: <功能描述> (#<PR号>) by @<作者>
+- ...
+
+## 🐛 问题修复 (Bug Fixes)
+
+- **<模块>**: <修复描述> (#<Issue号>) by @<修复者>
+- ...
+
+## 🔧 改进优化 (Improvements)
+
+- **<模块>**: <改进描述> (#<PR号>)
+- ...
+
+## 📚 文档更新 (Documentation)
+
+- <文档变更描述>
+- ...
+
+## 🧪 测试 (Tests)
+
+- <新增/改进的测试描述>
+- ...
+
+## 🏗️ 构建/CI (Build & CI)
+
+- <构建或 CI 变更描述>
+- ...
+
+## ⚠️ 破坏性变更 (Breaking Changes)
+
+- <破坏性变更描述及迁移指南>
+> 如果没有破坏性变更，此节不显示
+
+## 🙏 贡献者
+
+感谢以下贡献者参与本版本开发：
+
+@contributor-a, @contributor-b, @contributor-c
+
+## 📊 变更统计
+
+| 类型 | 数量 |
+|------|------|
+| 新功能 | N |
+| Bug 修复 | N |
+| 改进优化 | N |
+| 文档更新 | N |
+| 测试 | N |
+| 合计 | N |
+```
+
+## Commit 分类规则
+
+AI 根据 commit message 的关键词自动分类：
+
+| 分类 | 关键词前缀 | 示例 |
+|------|-----------|------|
+| ✨ 新功能 | `feat`, `feature`, `add`, `新增`, `支持` | `feat: add wiki module` |
+| 🐛 Bug 修复 | `fix`, `bugfix`, `hotfix`, `修复`, `解决` | `fix: resolve wiki encoding issue` |
+| 🔧 改进优化 | `refactor`, `perf`, `improve`, `优化`, `增强`, `完善`, `enhance` | `refactor: simplify batch logic` |
+| 📚 文档更新 | `docs`, `doc`, `文档`, `README`, `注释` | `docs: update API reference` |
+| 🧪 测试 | `test`, `tests`, `测试` | `test: add wiki module tests` |
+| 🏗️ 构建/CI | `build`, `ci`, `chore`, `构建`, `部署`, `Docker` | `ci: add FreeBSD build target` |
+| 💄 样式 | `style`, `fmt`, `格式化` | `style: fix indentation`（归入改进优化） |
+| ⚠️ 破坏性变更 | commit body 含 `BREAKING CHANGE` 或标题含 `!` | `feat!: change API response format` |
+
+> 完整分类规则和优先级逻辑见 [`REFERENCE.md`](REFERENCE.md)。
+
+## 使用场景
+
+### 场景 1：为指定版本生成 Release Notes
+
+```
+用户："为 v1.2.0 生成 Release Notes"
+
+AI 执行流程：
+1. release +list → 找到上一个版本 v1.1.0 的发布时间
+2. commit +list → 筛选 v1.1.0 之后的所有提交
+3. pr +list --state merged → 筛选同一时间段的已合并 PR
+4. issue +list --state closed → 筛选已关闭的 Issue
+5. 按分类规则整理 → 生成结构化 Release Notes
+```
+
+### 场景 2：生成完整 Changelog（所有版本）
+
+```
+用户："生成项目完整 Changelog"
+
+AI 执行流程：
+1. release +list → 获取所有版本
+2. 对每个版本区间分别采集 commit/PR/Issue
+3. 按版本倒序排列，生成完整 Changelog
+```
+
+### 场景 3：生成并一键发布
+
+```
+用户："生成 v1.2.0 Release Notes 并发布"
+
+AI 执行流程：
+1-5. 同场景 1，生成 Release Notes
+6. 确认用户意图："即将发布 v1.2.0 Release，内容如上，确认发布？"
+7. 用户确认后执行：
+   gitlink-cli release +create --owner <owner> --repo <repo> \
+     --tag v1.2.0 --name "v1.2.0" --body "<Release Notes 内容>"
+```
+
+## 发布命令
+
+```bash
+# 创建 Release（需要认证）
+gitlink-cli release +create --owner <owner> --repo <repo> \
+  --tag v1.2.0 \
+  --name "v1.2.0" \
+  --body "## ✨ 新功能\n- feat: xxx\n\n## 🐛 Bug 修复\n- fix: yyy"
+
+# 查看 Release 是否创建成功
+gitlink-cli release +view --owner <owner> --repo <repo> --id <version_id>
+```
+
+## 最佳实践
+
+- 所有数据采集命令使用 `--format json`
+- commit message 建议遵循 [Conventional Commits](https://www.conventionalcommits.org/) 规范，便于自动分类
+- 发布前让用户预览 Release Notes 并确认
+- 保留上一个版本的时间戳作为基线，避免遗漏或重复
+- 如果项目未发布过 Release，则从项目创建时间开始采集所有数据
+- 关联 PR 和 Issue 编号，增强 Release Notes 的可追溯性
+
+## 详细参考
+
+详见 [`REFERENCE.md`](REFERENCE.md) 了解完整的 commit 分类规则、数据字段映射和 API 注意事项。

skills/gitlink-changelog/examples/generate-release-notes.md

@@ -0,0 +1,112 @@
+# 示例：Release Notes 生成（真实数据）
+
+> 本示例基于 `chroe/gitlink-cli` 项目于 2026-06-09 在 Claude Code 中实际执行。
+> 展示从 v0.1.13-freebsd（上一个 Release，2026-06-03 23:51）到当前时间段的变更。
+
+---
+
+## 执行命令序列
+
+```bash
+# Step 1: 获取上一个版本（确定时间基线）
+gitlink-cli release +list --owner chroe --repo gitlink-cli --format json
+# 结果: 1 个版本, tag=v0.1.13-freebsd, created=2026-06-03 23:51
+
+# Step 2: 获取提交历史（AI 按时间戳筛选基线之后的提交）
+gitlink-cli commit +list --owner chroe --repo gitlink-cli --format json
+# 结果: 本页 20 个提交，基线之后 20 个（含 2 个 merge commit 需跳过）
+# 如需更多: gitlink-cli commit +list --owner chroe --repo gitlink-cli --page 2 --format json
+
+# Step 3: 获取 PR（本项目为 Fork，无 PR 记录）
+gitlink-cli pr +list --owner chroe --repo gitlink-cli --state merged --format json
+# 结果: merged_issues_size=0
+
+# Step 4: 获取已关闭 Issue
+gitlink-cli issue +list --owner chroe --repo gitlink-cli --state closed --format json
+# 结果: 6 个已关闭 Issue（#1~#6）
+```
+
+## Commit 分类过程（20 条提交）
+
+| # | commit message（首行） | 作者 | 分类结果 |
+|---|----------------------|------|---------|
+| 1 | `fix(skills): gitlink-health 评分算法修复...` | chroe | 🐛 Bug 修复 |
+| 2 | `feat(skills): 新增 3 个 AI Agent Skill...` | chroe | ✨ 新功能 |
+| 3 | `fix(showcase): 移除 rebase 残留...` | chroe | 🐛 Bug 修复 |
+| 4 | `refactor(showcase): 成员管理卡片优化` | chroe | 🔧 改进优化 |
+| 5 | `删除不可用的批量操作...新增批量评论...` | yetja | 🔧 改进优化 |
+| 6 | `refactor(showcase): 批量组织邀请合并...` | chroe | 🔧 改进优化 |
+| 7 | `fix(org): batch-invite 用户名自动解析...` | chroe | 🐛 Bug 修复 |
+| 8 | `chore: batch-delete 默认改用 fork 后的路径` | chroe | 🏗️ 构建/CI |
+| 9 | `chore: 更新仓库管理默认示例为真实仓库` | chroe | 🏗️ 构建/CI |
+| 10 | `chore: 恢复 repo 批量管理卡片` | chroe | 🏗️ 构建/CI |
+| 11 | `fix: 移除 batch-milestone (v1 PATCH不生效)` | chroe | 🐛 Bug 修复 |
+| 12 | `fix: 移除 batch-label (API不支持标签关联)` | chroe | 🐛 Bug 修复 |
+| 13 | `fix: 移除 assign/batch-assign (API不支持)` | chroe | 🐛 Bug 修复 |
+| 14 | `fix: assign 传递数字 ID + 保留 status_id` | chroe | 🐛 Bug 修复 |
+| 15 | `Merge branch 'master' of...` | chroe | ⏭️ 跳过 |
+| 16 | `feat: 添加 reopen/batch-reopen 命令` | chroe | ✨ 新功能 |
+| 17 | `delete test file` | chroe | ⏭️ 跳过 |
+| 18 | `add test file` | chroe | ⏭️ 跳过 |
+| 19 | `fix: watch/star 模块 resolveProjectID...` | chroe | 🐛 Bug 修复 |
+| 20 | `Merge branch 'master' of...` | chroe | ⏭️ 跳过 |
+
+## 生成的 Release Notes
+
+```markdown
+# 🚀 v0.2.0 (2026-06-09)
+
+> 基于 v0.1.13-freebsd (2026-06-03 23:51) 之后的变更生成
+
+## 📝 版本摘要
+
+本版本为 Skills 生态扩展版本，新增 3 个 AI Agent Skill（项目健康度报告、Release Notes 自动生成、Issue 智能分拣），
+新增 reopen/batch-reopen 命令，清理了多个 GitLink API 不支持的批量操作命令，并修复了 Showcase 前端和模块兼容性问题。
+
+## ✨ 新功能 (Features)
+
+- **skills**: 新增 `gitlink-health` Skill — 项目健康度报告（Issue/PR/贡献者/活跃度四维评分）(by @chroe)
+- **skills**: 新增 `gitlink-changelog` Skill — 从 commit/PR/Issue 历史自动生成 Release Notes (by @chroe)
+- **skills**: 新增 `gitlink-triage` Skill — Issue 智能分拣 + 新人引导 (by @chroe)
+- **issue**: 新增 reopen/batch-reopen 命令 (by @chroe)
+
+## 🐛 问题修复 (Bug Fixes)
+
+- **skills**: 修复 gitlink-health 评分算法 — fallback 策略和贡献集中度计算 (by @chroe)
+- **showcase**: 修复 rebase 残留 commit message 导致 JS 语法错误白屏 (by @chroe)
+- **org**: batch-invite 用户名自动解析为数字 ID，修正 Showcase 默认值 (by @chroe)
+- **watch/star**: 修复 resolveProjectID 兼容 project_id 字段 (by @chroe)
+- **batch**: 移除 batch-milestone（GitLink v1 PATCH 实际不生效）(by @chroe)
+- **batch**: 移除 batch-label（GitLink API 不支持 Issue 标签关联）(by @chroe)
+- **batch**: 移除 assign/batch-assign（GitLink API 不支持）(by @chroe)
+- **assign**: 传递数字 ID + 保留 status_id 防止状态被重置 (by @chroe)
+
+## 🔧 改进优化 (Improvements)
+
+- **showcase**: 成员管理卡片优化 — 组织 ID 查找改为 `org +info`，新增仓库选择器 (by @chroe)
+- **showcase**: 批量组织邀请合并到成员管理卡片 (by @chroe)
+- **batch**: 移除不可用的批量操作，新增批量评论、批量仓库管理 (by @yetja)
+
+## 🏗️ 构建/CI (Build & CI)
+
+- batch-delete 默认改用 fork 后的仓库路径 (by @chroe)
+- 更新仓库管理默认示例为真实仓库 (by @chroe)
+- 恢复 repo 批量管理卡片 (by @chroe)
+
+## 🙏 贡献者
+
+感谢以下贡献者参与本版本开发：
+
+@chroe, @yetja, @caoweiqiong (CWQ)
+
+## 📊 变更统计
+
+| 类型 | 数量 |
+|------|------|
+| 新功能 | 4 |
+| Bug 修复 | 8 |
+| 改进优化 | 3 |
+| 构建/CI | 3 |
+| 跳过（merge/test） | 4 |
+| 合计 | 18（有效 16）|
+```

skills/gitlink-ci/REFERENCE.md

@@ -0,0 +1,198 @@
+# gitlink-ci 参考手册
+
+> 本文档定义 CI/CD 的 API 字段映射、错误诊断模式、流水线配置参考和已知问题。
+
+---
+
+## 一、前置条件：DevOps 状态
+
+### 检查方法
+
+```bash
+gitlink-cli repo +info --owner <owner> --repo <repo> --format json
+# 关注字段: "open_devops": true/false
+```
+
+| open_devops | CI 命令可用性 |
+|-------------|-------------|
+| `true` | `ci +builds/logs/restart/stop` 全部可用 |
+| `false` | 所有 CI 命令返回 `[-1] 接口数据异常` |
+
+### 激活 DevOps
+
+```bash
+# 通过 Raw API 激活
+gitlink-cli api POST /<owner>/<repo>/activate
+
+# 或通过 Web 页面：
+# 仓库 → 设置 → DevOps → 开启
+```
+
+### 实测数据
+
+在 `chroe/gitlink-cli` 上（`open_devops: false`）：
+```bash
+gitlink-cli ci +builds --owner chroe --repo gitlink-cli
+# 输出: 错误: 获取构建列表失败: [-1] 接口数据异常
+```
+
+---
+
+## 二、CI API 字段映射
+
+### ci +builds 预期响应结构
+
+```json
+{
+  "ok": true,
+  "data": {
+    "builds": [
+      {
+        "id": 42,
+        "number": 42,
+        "status": "failed",
+        "branch": "feature/new-auth",
+        "commit": "abc1234def",
+        "commit_message": "feat: add new auth",
+        "created_at": "2026-06-12T10:00:00+08:00",
+        "duration": 120,
+        "stages": [
+          { "number": 1, "name": "build", "status": "success" },
+          { "number": 2, "name": "test", "status": "failed" }
+        ]
+      }
+    ],
+    "total_count": 50
+  }
+}
+```
+
+| 字段 | 用途 |
+|------|------|
+| `number` | 构建编号，用于 `ci +logs` / `ci +restart` / `ci +stop` |
+| `status` | `success` / `failed` / `running` / `stopped` |
+| `branch` | 触发构建的分支 |
+| `commit` | 触发构建的 commit SHA |
+| `stages` | 流水线阶段列表，每个 stage 有独立状态 |
+| `duration` | 构建耗时（秒） |
+
+### ci +logs API
+
+```
+GET /<owner>/<repo>/builds/<build-number>/logs/<stage>/<step>
+```
+
+参数说明：
+- `build-number`：从 `ci +builds` 获取
+- `stage`：阶段编号，默认 1（编译阶段）
+- `step`：步骤编号，默认 1
+
+---
+
+## 三、构建错误诊断模式库
+
+### 编译错误
+
+| 模式 | 正则 | 常见原因 |
+|------|------|---------|
+| 依赖缺失 | `cannot find package` | `go.mod` 或 `package.json` 不完整 |
+| 语法错误 | `syntax error` | 代码语法问题 |
+| 类型错误 | `cannot use .* as type` | 类型不匹配 |
+| 未定义引用 | `undefined: ` | 缺少 import 或拼写错误 |
+| 导入路径错误 | `no required module provides package` | Go module 路径变更 |
+
+### 测试失败
+
+| 模式 | 常见原因 |
+|------|---------|
+| `FAIL: TestXxx` | 测试断言失败 |
+| `panic: runtime error` | 测试中空指针或越界 |
+| `--- FAIL: TestXxx (0.00s)` | 测试立即失败（setup 错误） |
+| `too many arguments` | 测试函数签名不匹配 |
+
+### 环境/基础设施
+
+| 模式 | 常见原因 |
+|------|---------|
+| `connection refused` | 数据库/外部服务不可用 |
+| `out of memory` | 构建内存不足 |
+| `permission denied` | 密钥或文件权限问题 |
+| `docker: not found` | 构建环境缺少 Docker |
+| `No space left on device` | 磁盘空间不足 |
+
+### GitLink 特定错误
+
+| 模式 | 说明 |
+|------|------|
+| `[-1] 接口数据异常` | 仓库未开启 DevOps |
+| 返回 HTML 而非 JSON | API 路径错误（如缺少 `/v1/` 前缀） |
+
+---
+
+## 四、流水线配置文件
+
+### 文件位置
+
+```
+仓库根目录/
+└── .devops/
+    └── <流水线名称>.yml
+```
+
+### 示例配置（Go 项目）
+
+```yaml
+name: 构建部署
+on:
+  push:
+    branches: [master]
+jobs:
+  build:
+    runs-on: docker
+    steps:
+      - name: 构建
+        run: go build -o app .
+      - name: 测试
+        run: go test ./...
+      - name: 部署
+        run: |
+          ssh root@server "cd /opt/app && git pull && docker build -t app . && docker-compose up -d"
+```
+
+### 关键注意事项
+
+- Docker 构建需配置 `GOPROXY=https://goproxy.cn,direct`（国内网络）
+- 服务器 Docker daemon 需配置国内镜像加速器（`/etc/docker/daemon.json`）
+- GitLink 密钥管理：敏感信息通过 `deploy_server.server_password` 注入
+- 使用 `git fetch + git reset --hard` 替代 `git pull` 避免本地修改冲突
+
+---
+
+## 五、已知限制
+
+| 限制 | 说明 |
+|------|------|
+| DevOps 默认关闭 | 大部分仓库的 `open_devops` 为 `false`，需手动开启 |
+| 接口数据异常 | 通用错误码 `-1`，无结构化错误信息 |
+| 无构建触发 API | 无法通过 CLI 触发新构建，只能通过 git push 触发 |
+| 日志可能截断 | 长日志可能被分页或截断 |
+
+---
+
+## 六、常见问题
+
+### Q: 所有 CI 命令都返回"接口数据异常"？
+
+A: 99% 的情况是因为仓库未开启 DevOps。检查 `repo +info` 中的 `open_devops` 字段。
+
+### Q: 如何触发一次新构建？
+
+A: GitLink 没有"手动触发构建"的 API。只能通过 `git push` 到触发分支（如 master）来启动构建。
+
+### Q: ci +logs 的 stage/step 是什么意思？
+
+A: 每个流水线有多个 stage（阶段），每个 stage 有多个 step（步骤）。默认 stage=1, step=1 通常是第一个编译步骤。
+
+### Q: CI 构建没有日志输出？
+
+A: 尝试不同的 stage/step 组合。如果 stage=1,step=1 无输出，试试 stage=2,step=1。

skills/gitlink-ci/SKILL.md

@@ -1,55 +1,232 @@
 ---
 name: gitlink-ci
-version: 1.0.0
-description: "CI/CD 操作：查看构建列表、构建日志、重启/停止构建。当用户需要操作 GitLink CI 时触发。"
+version: 2.0.0
+description: "CI/CD 构建诊断与监控：检查 DevOps 状态、诊断构建失败、分析 CI 日志、管理构建生命周期。当用户需要排查 CI 失败、监控构建状态、配置流水线时触发。"
 metadata:
   requires:
     bins: ["gitlink-cli"]
   cliHelp: "gitlink-cli ci --help"
 ---
 
-# gitlink-ci（CI/CD 操作）
+# gitlink-ci（CI/CD 构建诊断与监控）
 
 **CRITICAL — 开始前必须先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)，其中包含认证、权限处理和 API 注意事项。**
-**CRITICAL — 所有 Shortcuts 在执行写入/删除操作前，务必先确认用户意图。**
+**CRITICAL — CI 功能依赖仓库开启 DevOps（`open_devops: true`）。未开启的仓库所有 CI 命令会返回"接口数据异常"。使用前务必检查 DevOps 状态。**
+**CRITICAL — `ci +restart` 和 `ci +stop` 是写入操作，执行前需确认用户意图。**
 **CRITICAL — GitLink 操作只能用 `gitlink-cli`。禁止用 `gh`（GitHub CLI）操作 GitLink 资源。`gh` 仅适用于 GitHub 平台。**
 
 > **前置条件：** 先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md) 了解认证和全局参数。
 
-## Shortcuts
+## 概述
 
-| Shortcut | 说明 | 需要认证 |
-|----------|------|----------|
-| `ci +builds` | 构建列表 | 是 |
-| `ci +logs` | 构建日志 | 是 |
-| `ci +restart` | 重启构建 | 是 |
-| `ci +stop` | 停止构建 | 是 |
+本 Skill 通过 gitlink-cli 的 CI 命令和 Raw API，由 AI 分析后完成：
+1. **DevOps 状态检查**：扫描仓库是否开启 CI，未开启的引导激活
+2. **构建失败诊断**：获取失败构建的日志 → AI 分析错误 → 定位根因 → 建议修复
+3. **CI 健康监控**：查看近期构建成功率、平均构建时间
+4. **流水线配置审计**：检查 `.devops/` 目录下的流水线文件
 
-## 使用示例
+## 前置检查命令
+
+### 必须：确认 DevOps 状态
 
 ```bash
-# 查看构建列表
-gitlink-cli ci +builds --owner myuser --repo myrepo
-
-# 查看构建日志
-gitlink-cli ci +logs --build 42 --stage 1 --step 1
-
-# 重启构建
-gitlink-cli ci +restart --build 42
-
-# 停止构建
-gitlink-cli ci +stop --build 42
+# 检查仓库是否开启 DevOps
+gitlink-cli repo +info --owner <owner> --repo <repo> --format json | grep open_devops
 ```
 
-## Raw API 补充
+如果返回 `"open_devops": false`，后续 CI 命令将不可用。需要先激活：
 
 ```bash
-# 激活 CI
-gitlink-cli api POST /:owner/:repo/activate
+# 激活 DevOps
+gitlink-cli api POST /<owner>/<repo>/activate
+```
 
+> ⚠️ **实测**：`chroe/gitlink-cli` 等仓库的 `open_devops` 为 `false`，调用 `ci +builds` 会返回"接口数据异常"。这不是 CLI 的 bug，而是 GitLink 平台的前置要求。
+
+## 数据采集命令
+
+### 第一步：获取构建列表
+
+```bash
+# 查看构建历史
+gitlink-cli ci +builds --owner <owner> --repo <repo> --format json
+
+# 分页
+gitlink-cli ci +builds --owner <owner> --repo <repo> --page 2 --format json
+```
+
+### 第二步：查看失败构建日志
+
+```bash
+# 查看默认 stage=1, step=1 的日志
+gitlink-cli ci +logs --owner <owner> --repo <repo> --build <build-number>
+
+# 查看特定 stage/step（如 stage 2 = 测试阶段）
+gitlink-cli ci +logs --owner <owner> --repo <repo> --build <build-number> --stage 2 --step 1
+```
+
+### 第三步：查看 CI 配置
+
+```bash
+# 检查 CI 授权状态
+gitlink-cli api GET /<owner>/<repo>/ci_authorize
+
+# 查看构建详情
+gitlink-cli api GET /<owner>/<repo>/builds/<build_id>
+```
+
+## AI 分析规则
+
+### 构建失败诊断流程
+
+```
+1. ci +builds → 筛选 status=failed 的构建
+2. ci +logs --build <n> → 获取日志
+3. AI 分析日志中的错误模式
+4. 匹配已知错误类型 → 给出修复建议
+5. 如有多个 stage，逐 stage 排查
+```
+
+### 常见错误模式识别
+
+| 日志关键词 | 诊断 | 建议修复 |
+|-----------|------|---------|
+| `cannot find package` | 依赖缺失 | 检查 `go.mod` 或 `package.json` |
+| `syntax error` / `unexpected token` | 语法错误 | 检查最近提交的代码 |
+| `permission denied` | 权限不足 | 检查密钥配置、文件权限 |
+| `connection refused` | 服务不可达 | 检查外部服务/数据库连接 |
+| `out of memory` / `killed` | 资源不足 | 优化内存使用或增加构建资源 |
+| `No such file` | 文件缺失 | 检查 `.devops/` 配置文件路径 |
+| `docker: command not found` | 环境缺失 | 确认构建环境有 Docker |
+| `exit status 1` / `FAIL` | 测试失败 | 查看具体测试输出 |
+| `timeout` | 构建超时 | 优化构建脚本或增加超时时间 |
+
+### 构建健康评分
+
+| 指标 | 计算方式 |
+|------|---------|
+| 成功率 | 成功构建 / 总构建 × 100% |
+| 平均耗时 | 所有构建的 `duration` 平均值 |
+| 失败趋势 | 最近 10 次构建中失败次数的变化方向 |
+
+## 执行命令
+
+### 重启失败的构建
+
+```bash
+gitlink-cli ci +restart --owner <owner> --repo <repo> --build <build-number>
+```
+
+### 停止异常构建
+
+```bash
+gitlink-cli ci +stop --owner <owner> --repo <repo> --build <build-number>
+```
+
+### 管理 CI 状态
+
+```bash
 # 停用 CI
-gitlink-cli api DELETE /:owner/:repo/deactivate
+gitlink-cli api DELETE /<owner>/<repo>/deactivate
 
-# CI 授权状态
-gitlink-cli api GET /:owner/:repo/ci_authorize
+# 激活 CI
+gitlink-cli api POST /<owner>/<repo>/activate
 ```
+
+## 输出格式
+
+### 构建失败诊断报告
+
+```markdown
+# 🔧 构建失败诊断报告
+
+> 仓库：<owner/repo>
+> 构建编号：#<n>
+> 诊断时间：<timestamp>
+
+## 失败概况
+
+| 项目 | 详情 |
+|------|------|
+| 构建编号 | #42 |
+| 触发分支 | feature/new-auth |
+| 失败 Stage | stage 2 (测试) |
+| 失败时间 | 2026-06-12 10:30 |
+
+## 错误日志（关键部分）
+
+```
+[ERROR] cannot find package "github.com/example/lib"
+  at main.go:5
+```
+
+## AI 诊断
+
+| 错误类型 | 依赖缺失 |
+|---------|---------|
+| **根因** | `go.mod` 中缺少 `github.com/example/lib` 依赖 |
+| **影响范围** | 所有 import 该包的 Go 文件 |
+| **修复建议** | 运行 `go get github.com/example/lib && go mod tidy` 后提交 |
+
+## 修复步骤
+
+1. 本地运行 `go get github.com/example/lib`
+2. 运行 `go mod tidy` 更新依赖
+3. 提交修改后的 `go.mod` 和 `go.sum`
+4. 运行 `gitlink-cli ci +restart --build 42` 重启构建
+```
+
+## 使用场景
+
+### 场景 1：构建失败排查
+
+```
+用户："帮我看看 repo 的构建为什么失败了"
+
+AI 执行流程：
+1. repo +info → 检查 open_devops
+2. ci +builds → 获取构建列表，找到失败构建
+3. ci +logs --build <n> → 获取失败日志
+4. AI 分析日志中的错误模式
+5. 匹配已知错误 → 生成诊断报告和修复建议
+6. 询问用户是否要重启构建
+```
+
+### 场景 2：CI 配置审计
+
+```
+用户："帮我检查项目 CI 配置是否正常"
+
+AI 执行流程：
+1. repo +info → 检查 open_devops
+2. ci +builds → 查看最近的构建记录
+3. raw api GET /<owner>/<repo>/ci_authorize → 检查授权状态
+4. 计算近期构建成功率
+5. 检查 .devops/ 目录下的流水线配置（如有本地代码）
+6. 生成 CI 健康度报告
+```
+
+### 场景 3：批量 DevOps 检查
+
+```
+用户："检查我所有仓库的 DevOps 开启情况"
+
+AI 执行流程：
+1. repo +list --category all → 获取所有仓库
+2. 筛选 open_devops = true 的仓库
+3. 对开启了 DevOps 的仓库，获取最近构建状态
+4. 对未开启的仓库，列出可激活的选项
+5. 生成 DevOps 覆盖率报告
+```
+
+## 最佳实践
+
+- **先查 DevOps 状态**：每次 CI 操作前检查 `open_devops`，避免"接口数据异常"错误
+- **逐 stage 排查**：多 stage 流水线中，从第一个失败的 stage 开始分析
+- **关联 commit**：失败构建通常与最近的代码变更相关，关联 `commit +list` 查看
+- **重启前确认**：确认已修复根因后再重启，避免反复失败
+- **日志截断**：CI 日志可能很长，AI 应提取错误关键词而非全文搬运
+
+## 详细参考
+
+详见 [`REFERENCE.md`](REFERENCE.md) 了解 CI API 字段映射、错误诊断模式和流水线配置。

skills/gitlink-ci/examples/ci-devops-check.md

@@ -0,0 +1,144 @@
+# 示例：CI DevOps 状态检查与构建诊断（真实数据）
+
+> 本示例基于真实仓库于 2026-06-12 在 Claude Code 中实际执行。
+> 展示 DevOps 状态检查 → 发现未开启 → 给出激活方案的完整流程。
+
+---
+
+## 场景：发现 CI 不可用
+
+用户想查看 `chroe/gitlink-cli` 的构建状态。
+
+### Step 1: 检查 DevOps 状态
+
+```bash
+gitlink-cli repo +info --owner chroe --repo gitlink-cli --format json | python3 -c "
+import sys,json
+d = json.load(sys.stdin)['data']
+print(f'DevOps: {d[\"open_devops\"]}')
+print(f'项目ID: {d[\"project_id\"]}')
+"
+```
+
+真实输出：
+```
+DevOps: False
+项目ID: 1547045
+```
+
+### Step 2: 尝试 CI 命令（预期失败）
+
+```bash
+gitlink-cli ci +builds --owner chroe --repo gitlink-cli
+```
+
+真实输出：
+```
+错误: 获取构建列表失败: [-1] 接口数据异常
+```
+
+### Step 3: 检查是否有其他仓库开启了 DevOps
+
+```bash
+gitlink-cli repo +list --category all --format json | python3 -c "
+import sys,json
+for p in json.load(sys.stdin)['data']['projects']:
+    if p.get('open_devops'):
+        print(f\"  ✅ {p['identifier']} (devops enabled)\")
+"
+```
+
+真实输出（在 yetja 用户下）：
+```
+（无输出 — 没有任何仓库开启 DevOps）
+```
+
+甚至 Gitlink 组织的热门仓库也是关闭的：
+```
+Gitlink/build        → devops: True   ← 仅有此仓库开启
+Gitlink/gitlink-cli  → devops: False
+Gitlink/forgeplus    → devops: False
+```
+
+## AI 分析
+
+```markdown
+# 🔧 CI DevOps 状态检查报告
+
+> 检查时间：2026-06-12
+> 检查范围：yetja 用户下的仓库 + Gitlink 组织热门仓库
+
+## 总览
+
+| 指标 | 数值 |
+|------|------|
+| 检查仓库数 | 20 |
+| DevOps 已开启 | 1 (5%) |
+| DevOps 未开启 | 19 (95%) |
+
+## DevOps 已开启的仓库
+
+| 仓库 | 说明 |
+|------|------|
+| Gitlink/build | GitLink 官方 CI/CD 项目 |
+
+## 关于 chroe/gitlink-cli
+
+- **DevOps 状态**：未开启（`open_devops: false`）
+- **CI 命令**：全部不可用，返回 `[-1] 接口数据异常`
+- **项目类型**：Fork 自 `Gitlink/gitlink-cli`
+- **项目 ID**：1547045
+
+## 如何开启 DevOps
+
+### 方法 1：Web 页面（推荐）
+
+1. 打开 https://gitlink.org.cn/chroe/gitlink-cli
+2. 进入 设置 → DevOps
+3. 点击"开启 DevOps"
+
+### 方法 2：CLI Raw API
+
+```bash
+gitlink-cli api POST /chroe/gitlink-cli/activate
+```
+
+> ⚠️ `api POST` 存在已知 URL 拼接 bug，可能返回 404。建议使用 Web 页面。
+
+## 开启后的流水线配置
+
+本项目已有 `.devops/构建部署Showcase.yml` 配置文件：
+
+- **触发条件**：push 到 master 分支
+- **构建环境**：Docker 容器
+- **部署目标**：118.31.4.168:9090
+- **容器名**：gitlink-cli-showcase
+
+开启 DevOps 后，每次 push 将自动触发构建和部署。
+```
+
+---
+
+## 补充示例：开启 DevOps 后的工作流（预期）
+
+如果 DevOps 已开启，典型的构建诊断流程如下：
+
+```bash
+# 1. 获取近期构建
+gitlink-cli ci +builds --owner chroe --repo gitlink-cli --format json
+
+# 2. 找到失败构建，查看日志
+gitlink-cli ci +logs --build 42 --stage 1 --step 1
+
+# 3. AI 分析日志中的错误
+# （日志内容示例）
+# [ERROR] cannot find package "github.com/example/lib"
+
+# 4. AI 诊断：依赖缺失 → 建议 go get + go mod tidy
+
+# 5. 修复后重启构建
+gitlink-cli ci +restart --build 42
+
+# 6. 验证修复
+gitlink-cli ci +logs --build 42 --stage 1 --step 1
+```