From 1ae613717bf4fb8a10c61ef5cd1decf5e465c623 Mon Sep 17 00:00:00 2001
From: Ct201314 <1195214305@qq.com>
Date: Sat, 6 Jun 2026 15:31:10 +0800
Subject: [PATCH 1/2] feat(skills): add gitlink-stale skill

---
 skills/gitlink-stale/SKILL.md                 | 82 +++++++++++++++++++
 .../gitlink-stale/references/api-reference.md | 55 +++++++++++++
 skills/gitlink-stale/references/grading.md    | 34 ++++++++
 3 files changed, 171 insertions(+)
 create mode 100644 skills/gitlink-stale/SKILL.md
 create mode 100644 skills/gitlink-stale/references/api-reference.md
 create mode 100644 skills/gitlink-stale/references/grading.md

diff --git a/skills/gitlink-stale/SKILL.md b/skills/gitlink-stale/SKILL.md
new file mode 100644
index 0000000..585720b
--- /dev/null
+++ b/skills/gitlink-stale/SKILL.md
@@ -0,0 +1,82 @@
+---
+name: gitlink-stale
+version: 1.0.0
+description: "陈旧 Issue/PR 清理：检测久未更新的开放 Issue 与 PR，按陈旧度分级（活跃/留意/陈旧/僵尸），生成清理建议清单。当用户提到「陈旧 Issue」「积压」「久未更新」「清理 Issue」「stale」「僵尸 PR」「待清理」时触发。"
+metadata:
+  requires:
+    bins: ["gitlink-cli"]
+    optional_bins: ["python"]
+  cliHelp: "gitlink-cli issue --help"
+---
+
+# gitlink-stale（陈旧 Issue/PR 清理）
+
+**CRITICAL — 开始前先阅读 [`../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。**
+**CRITICAL — 关闭 Issue/PR 属于写操作，执行前必须征得用户确认。本技能只生成建议清单，不自动关闭。**
+
+> **前置条件：** 先阅读 [`../gitlink-shared/SKILL.md`](../gitlink-shared/SKILL.md)。
+
+## 何时使用本技能
+
+- 维护者想清理长期无人跟进的 Issue/PR，控制积压
+- 用户问「有哪些 Issue 很久没动了」「该清理哪些 PR」
+- 定期做 backlog 健康检查
+
+## 何时不使用
+
+- 项目综合健康度 → 用 `gitlink-health` / `gitlink-insight`
+- 普通列出 Issue/PR → 用 `gitlink-issue` / `gitlink-pr`
+
+## 陈旧度分级
+
+| 等级 | 停滞时长 | 含义 |
+|:----:|:--------:|------|
+| 🟢 活跃 | ≤ 14 天 | 近期有更新 |
+| 🟡 留意 | ≤ 45 天 | 开始放缓 |
+| 🟠 陈旧 | ≤ 120 天 | 需要跟进 |
+| 🔴 僵尸 | > 120 天 | 建议评估关闭 |
+
+## 工作流：扫描陈旧项
+
+### 方式 A：配套脚本（推荐）
+
+```bash
+python scripts/stale.py --owner Gitlink --repo gitlink-cli
+python scripts/stale.py --owner Gitlink --repo gitlink-cli --format json
+```
+
+参数说明：
+
+| 参数 | 类型 | 必填 | 说明 |
+|------|------|:----:|------|
+| `--owner` / `--repo` | string | 是* | 仓库（*或用 `--slug`） |
+| `--slug` | string | 否 | `owner/repo` 或完整 URL |
+| `--format` | string | 否 | `markdown`（默认）或 `json` |
+| `--output` | string | 否 | 输出文件 |
+
+### 方式 B：用 gitlink-cli 命令
+
+```bash
+gitlink-cli issue +list --owner Gitlink --repo gitlink-cli --state open --format json
+gitlink-cli pr +list --owner Gitlink --repo gitlink-cli --state open --format json
+```
+
+### 处理建议（写操作，需确认）
+
+```bash
+# 确认后关闭陈旧 Issue（web 序号）
+gitlink-cli issue +close --owner Gitlink --repo gitlink-cli --number <web序号>
+```
+
+## API 注意事项
+
+- 停滞天数依据 `updated_at` / `created_at` 的相对时间（如"2个月前"）估算，为粗略值。
+- 只扫描开放项（Issue 状态非「关闭」、PR `pull_request_status=0`）。
+- 关闭操作是写操作，本技能只给建议，需用户确认后执行。
+
+## References
+
+- [api-reference.md](references/api-reference.md) — 采集接口与字段
+- [grading.md](references/grading.md) — 陈旧度分级与时间估算规则
+- [gitlink-shared](../gitlink-shared/SKILL.md) — 认证、全局参数、安全规则
diff --git a/skills/gitlink-stale/references/api-reference.md b/skills/gitlink-stale/references/api-reference.md
new file mode 100644
index 0000000..d5bbe73
--- /dev/null
+++ b/skills/gitlink-stale/references/api-reference.md
@@ -0,0 +1,55 @@
+# gitlink-stale API 参考
+
+> **前置条件：** 先阅读 [`../../gitlink-shared/SKILL.md`](../../gitlink-shared/SKILL.md)。
+
+本技能采集只读；关闭操作由维护者确认执行。
+
+## 采集的接口
+
+### Issue 列表
+
+```
+GET /:owner/:repo/issues.json?page=1&limit=50
+```
+
+使用字段：
+
+| 字段 | 用途 |
+|------|------|
+| `id` / `name` / `subject` | 编号与标题 |
+| `issue_status` | 过滤已关闭（含"关"/closed 的跳过） |
+| `updated_at` / `created_at` | 相对时间，估算停滞天数 |
+| `author_login` | 作者 |
+
+### PR 列表
+
+```
+GET /:owner/:repo/pulls.json?page=1&limit=50
+```
+
+| 字段 | 用途 |
+|------|------|
+| `pull_request_number` | 编号 |
+| `pull_request_status` | 仅取 0（open） |
+| `pr_time` / `updated_at` | 相对时间，估算停滞天数 |
+
+## 时间估算
+
+`updated_at`/`created_at` 多为相对时间字符串（"3天前"/"2个月前"/"1年前"），
+本技能用正则换算为天数：天/周(×7)/月(×30)/年(×365)，"刚刚/小时"记为 0。
+这是粗略估算，用于分级而非精确计时。
+
+## 写操作（关闭）
+
+```
+gitlink-cli issue +close --owner <o> --repo <r> --number <web序号>
+```
+
+属写操作，需用户确认。本技能只生成建议清单。
+
+## 输出字段（JSON）
+
+```json
+{"total_open": 27, "by_grade": {"活跃":20,"留意":6,"陈旧":1},
+ "cleanup_count": 1, "cleanup": [{"kind","id","title","age_days","grade","author"}]}
+```
diff --git a/skills/gitlink-stale/references/grading.md b/skills/gitlink-stale/references/grading.md
new file mode 100644
index 0000000..4c5ca46
--- /dev/null
+++ b/skills/gitlink-stale/references/grading.md
@@ -0,0 +1,34 @@
+# 陈旧度分级与时间估算规则
+
+## 分级阈值
+
+按"最近更新距今天数"分四级：
+
+| 等级 | 停滞天数 | emoji | 处理建议 |
+|:----:|:--------:|:-----:|----------|
+| 活跃 | ≤ 14 | 🟢 | 无需处理 |
+| 留意 | 15 – 45 | 🟡 | 关注进展 |
+| 陈旧 | 46 – 120 | 🟠 | 跟进或补充信息 |
+| 僵尸 | > 120 | 🔴 | 评估关闭 |
+| 未知 | 无法解析时间 | ⚪ | 人工核对 |
+
+阈值集中在 `scripts/stale.py` 的 `grade()` 函数，可按项目节奏调整。
+
+## 时间估算规则
+
+停滞天数来自 `updated_at`/`created_at` 的相对时间字符串：
+
+| 文本 | 换算 |
+|------|------|
+| 刚刚 / N 秒前 / N 分钟前 / N 小时前 | 0 天 |
+| N 天前 | N |
+| N 周前 | N × 7 |
+| N 个月前 / N月前 | N × 30 |
+| N 年前 | N × 365 |
+
+无法解析时记为"未知"，不计入清理建议。
+
+## 清理清单
+
+「陈旧」+「僵尸」项进入清理建议清单，按停滞天数降序，最多展示 30 条。
+报告同时给出分等级的处理建议与关闭命令示例。

From 8c6a00f4079d17b74b71de1e8d857aeded8ca2d8 Mon Sep 17 00:00:00 2001
From: Ct201314 <1195214305@qq.com>
Date: Fri, 12 Jun 2026 18:01:47 +0800
Subject: [PATCH 2/2] =?UTF-8?q?feat(skills):=20=E8=A1=A5=E5=85=85=20gitlin?=
 =?UTF-8?q?k-stale=20=E9=85=8D=E5=A5=97=E8=84=9A=E6=9C=AC=E4=B8=8E?=
 =?UTF-8?q?=E6=B5=8B=E8=AF=95?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

将 SKILL.md 方式A引用的脚本（scripts/）与单元测试（tests/）纳入 PR，使其自包含可运行。
---
 skills/gitlink-stale/scripts/glapi.py    | 241 +++++++++++++++++++++++
 skills/gitlink-stale/scripts/stale.py    | 194 ++++++++++++++++++
 skills/gitlink-stale/tests/test_stale.py |  86 ++++++++
 3 files changed, 521 insertions(+)
 create mode 100644 skills/gitlink-stale/scripts/glapi.py
 create mode 100644 skills/gitlink-stale/scripts/stale.py
 create mode 100644 skills/gitlink-stale/tests/test_stale.py

diff --git a/skills/gitlink-stale/scripts/glapi.py b/skills/gitlink-stale/scripts/glapi.py
new file mode 100644
index 0000000..41db09d
--- /dev/null
+++ b/skills/gitlink-stale/scripts/glapi.py
@@ -0,0 +1,241 @@
+"""GitLink 公开 API 共享客户端。
+
+供 gitlink-skills-pack 下各 Skill 的脚本复用。仅依赖 Python 标准库，
+无需第三方包，便于在受限环境或 Agent 沙箱中运行。
+
+数据全部来自 GitLink 平台公开接口（https://www.gitlink.org.cn/api），
+默认无需 token；如需访问私有仓库，可传入 token。
+
+所有方法均为只读，不修改任何远程数据。
+"""
+
+from __future__ import annotations
+
+import base64
+import json
+import time
+import urllib.error
+import urllib.parse
+import urllib.request
+from pathlib import Path
+from typing import Any
+
+API_BASE = "https://www.gitlink.org.cn/api"
+USER_AGENT = "gitlink-skills-pack/1.0 (+https://www.gitlink.org.cn)"
+DEFAULT_TIMEOUT = 30
+COMMIT_PAGE_SIZE = 50  # GitLink commits 接口每页硬上限
+
+
+class GitLinkError(RuntimeError):
+    """API 调用中不可恢复的错误。"""
+
+
+class GitLinkClient:
+    """GitLink 公开数据接口客户端。
+
+    带可选文件缓存：同一资源重复读取不重复打网，对平台友好。
+    """
+
+    def __init__(self, base: str = API_BASE, token: str | None = None,
+                 timeout: int = DEFAULT_TIMEOUT, cache_dir: Path | None = None) -> None:
+        self.base = base.rstrip("/")
+        self.token = token
+        self.timeout = timeout
+        self.cache_dir = cache_dir
+        if self.cache_dir:
+            self.cache_dir.mkdir(parents=True, exist_ok=True)
+
+    # ------------------------------------------------------------------
+    # 底层请求
+    # ------------------------------------------------------------------
+    def _cache_path(self, url: str) -> Path | None:
+        if not self.cache_dir:
+            return None
+        safe = urllib.parse.quote(url, safe="")
+        return self.cache_dir / f"{safe}.json"
+
+    def get(self, path: str, query: dict[str, Any] | None = None) -> Any:
+        """GET 请求，返回解析后的 JSON（dict/list）或 None。"""
+        url = f"{self.base}/{path.lstrip('/')}"
+        if query:
+            url = f"{url}?{urllib.parse.urlencode(query)}"
+
+        cache_path = self._cache_path(url)
+        if cache_path and cache_path.exists():
+            return json.loads(cache_path.read_text(encoding="utf-8"))
+
+        headers = {"Accept": "application/json", "User-Agent": USER_AGENT}
+        if self.token:
+            headers["Authorization"] = f"Bearer {self.token}"
+
+        req = urllib.request.Request(url, headers=headers)
+        try:
+            with urllib.request.urlopen(req, timeout=self.timeout) as resp:
+                raw = resp.read().decode("utf-8", errors="replace")
+        except urllib.error.HTTPError as exc:
+            raise GitLinkError(f"HTTP {exc.code}: {url}") from exc
+        except urllib.error.URLError as exc:
+            raise GitLinkError(f"网络错误: {url} -> {exc.reason}") from exc
+
+        text = raw.strip()
+        if not text or text in ("null", "{}", "[]"):
+            data: Any = None
+        elif text[0] in "{[":
+            try:
+                data = json.loads(text)
+            except json.JSONDecodeError as exc:
+                raise GitLinkError(f"响应非 JSON: {url}") from exc
+        else:
+            raise GitLinkError(f"响应非 JSON（可能是 HTML）: {url}")
+
+        if cache_path is not None:
+            cache_path.write_text(json.dumps(data, ensure_ascii=False), encoding="utf-8")
+        return data
+
+    # ------------------------------------------------------------------
+    # 资源访问（高层封装）
+    # ------------------------------------------------------------------
+    def repo_info(self, owner: str, repo: str) -> dict[str, Any]:
+        """仓库元信息。"""
+        data = self.get(f"{owner}/{repo}.json")
+        return data if isinstance(data, dict) else {}
+
+    def issues(self, owner: str, repo: str, limit: int = 50,
+               page: int = 1) -> list[dict[str, Any]]:
+        """Issue 列表。"""
+        data = self.get(f"{owner}/{repo}/issues.json", {"page": page, "limit": limit})
+        return _extract_list(data, ("issues",))
+
+    def issue_detail(self, owner: str, repo: str, number: int) -> dict[str, Any]:
+        """单个 Issue 详情（含完整字段）。"""
+        data = self.get(f"{owner}/{repo}/issues/{number}.json")
+        return data if isinstance(data, dict) else {}
+
+    def pulls(self, owner: str, repo: str, limit: int = 50,
+              page: int = 1) -> list[dict[str, Any]]:
+        """PR 列表。"""
+        data = self.get(f"{owner}/{repo}/pulls.json", {"page": page, "limit": limit})
+        return _extract_list(data, ("issues", "pulls"))
+
+    def contributors(self, owner: str, repo: str) -> list[dict[str, Any]]:
+        """贡献者列表。"""
+        data = self.get(f"{owner}/{repo}/contributors.json")
+        return _extract_list(data, ("list",))
+
+    def commits(self, owner: str, repo: str, max_pages: int = 4) -> list[dict[str, Any]]:
+        """提交列表（按需翻页，每页 50 条，以 total_count 为终止依据）。"""
+        out: list[dict[str, Any]] = []
+        total: int | None = None
+        for page in range(1, max(1, max_pages) + 1):
+            data = self.get(f"{owner}/{repo}/commits.json",
+                            {"page": page, "limit": COMMIT_PAGE_SIZE})
+            if total is None and isinstance(data, dict):
+                total = _safe_int(data.get("total_count")) or None
+            page_items = _extract_list(data, ("commits",))
+            if not page_items:
+                break
+            out.extend(page_items)
+            if total is not None and len(out) >= total:
+                break
+        return out
+
+    def list_dir(self, owner: str, repo: str, path: str = "",
+                 ref: str = "master") -> list[dict[str, Any]]:
+        """列出目录下的条目（文件与子目录）。
+
+        返回的每个 entry 含 name / path / type(file|dir) / sha / size，
+        文件类型的 entry 还可能直接带明文 content。
+        """
+        data = self.get(f"{owner}/{repo}/sub_entries.json",
+                        {"filepath": path, "ref": ref})
+        # 查询目录时 entries 为 list；查询单文件时 entries 为单个 dict。
+        # 统一归一化为 list，便于下游处理。
+        if isinstance(data, dict):
+            entries = data.get("entries")
+            if isinstance(entries, dict):
+                return [entries]
+            if isinstance(entries, list):
+                return entries
+        return _extract_list(data, ("entries",))
+
+    def file_content(self, owner: str, repo: str, filepath: str,
+                     ref: str = "master") -> str | None:
+        """读取单个文件的文本内容。
+
+        GitLink 的 sub_entries 接口对单文件查询会在 entries 中返回明文 content，
+        据此取出。文件不存在或无内容时返回 None。
+        """
+        entries = self.list_dir(owner, repo, filepath, ref)
+        target = filepath.rsplit("/", 1)[-1]
+        for entry in entries:
+            if entry.get("type") == "file" and entry.get("name") == target:
+                content = entry.get("content")
+                if isinstance(content, str):
+                    return content
+        # 回退：部分情况下单文件查询 entries 仅一项
+        if len(entries) == 1 and entries[0].get("type") == "file":
+            content = entries[0].get("content")
+            if isinstance(content, str):
+                return content
+        return None
+
+    def readme(self, owner: str, repo: str, ref: str = "master") -> str | None:
+        """读取仓库 README（自动 base64 解码）。"""
+        data = self.get(f"{owner}/{repo}/readme.json", {"ref": ref})
+        if not isinstance(data, dict):
+            return None
+        content = data.get("content")
+        if not isinstance(content, str):
+            return None
+        # 注意：GitLink 的 readme.json 虽然 encoding 标为 base64，
+        # 实测 content 多为明文 Markdown。先探测明文特征，命中则直接返回；
+        # 否则再尝试 base64 解码。
+        stripped = content.lstrip()
+        if stripped.startswith(("#", "<", "[", "-", "*", "本", "这", "项")) or "\n" in content[:200]:
+            return content
+        try:
+            raw = base64.b64decode(content.encode("ascii", "ignore"))
+            decoded = raw.decode("utf-8", errors="replace")
+            # 解码结果若不像文本（大量替换符），回退为原文
+            if decoded.count("\ufffd") > len(decoded) * 0.1:
+                return content
+            return decoded
+        except (ValueError, TypeError):
+            return content
+
+
+# ----------------------------------------------------------------------------
+# 辅助
+# ----------------------------------------------------------------------------
+
+def _extract_list(payload: Any, keys: tuple[str, ...]) -> list[Any]:
+    """从可能嵌套的响应中提取第一个匹配键的列表。"""
+    if isinstance(payload, list):
+        return payload
+    if isinstance(payload, dict):
+        for key in keys:
+            value = payload.get(key)
+            if isinstance(value, list):
+                return value
+    return []
+
+
+def _safe_int(value: Any, default: int = 0) -> int:
+    try:
+        return int(value)
+    except (TypeError, ValueError):
+        return default
+
+
+def split_owner_repo(slug: str) -> tuple[str, str]:
+    """把 'owner/repo' 或完整 URL 解析为 (owner, repo)。"""
+    s = slug.strip()
+    if s.startswith("http"):
+        parts = urllib.parse.urlparse(s).path.strip("/").split("/")
+        if len(parts) >= 2:
+            return parts[0], parts[1].replace(".git", "")
+        raise GitLinkError(f"无法从 URL 解析 owner/repo: {slug}")
+    if "/" in s:
+        owner, repo = s.split("/", 1)
+        return owner, repo.replace(".git", "")
+    raise GitLinkError(f"格式应为 owner/repo: {slug}")
diff --git a/skills/gitlink-stale/scripts/stale.py b/skills/gitlink-stale/scripts/stale.py
new file mode 100644
index 0000000..b4eb0ed
--- /dev/null
+++ b/skills/gitlink-stale/scripts/stale.py
@@ -0,0 +1,194 @@
+"""gitlink-stale：陈旧 Issue/PR 清理。
+
+检测仓库中久未更新的开放 Issue 与 PR，按陈旧程度分级（活跃/留意/陈旧/僵尸），
+生成清理建议清单，帮助维护者控制积压。
+
+数据来自 GitLink 公开 API（只读），无需登录。生成的只是建议清单，
+是否关闭由维护者通过 gitlink-cli 自行决定。
+
+用法：
+    python stale.py --owner Gitlink --repo gitlink-cli
+    python stale.py --owner Gitlink --repo gitlink-cli --format json
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import re
+import sys
+from pathlib import Path
+from typing import Any
+
+sys.path.insert(0, str(Path(__file__).resolve().parent))
+from glapi import GitLinkClient, GitLinkError, split_owner_repo
+
+if hasattr(sys.stdout, "reconfigure"):
+    try:
+        sys.stdout.reconfigure(encoding="utf-8")
+    except Exception:
+        pass
+
+
+def parse_relative_days(text: str) -> int | None:
+    """把相对时间（如 "3天前"/"2个月前"/"1年前"/"刚刚"）粗略换算成天数。"""
+    if not text:
+        return None
+    t = str(text).strip()
+    if "刚" in t or "分钟" in t or "秒" in t or "小时" in t:
+        return 0
+    m = re.search(r"(\d+)\s*天", t)
+    if m:
+        return int(m.group(1))
+    m = re.search(r"(\d+)\s*周", t)
+    if m:
+        return int(m.group(1)) * 7
+    m = re.search(r"(\d+)\s*个?月", t)
+    if m:
+        return int(m.group(1)) * 30
+    m = re.search(r"(\d+)\s*年", t)
+    if m:
+        return int(m.group(1)) * 365
+    return None
+
+
+# 陈旧度分级阈值（天）
+def grade(days: int | None) -> str:
+    if days is None:
+        return "未知"
+    if days <= 14:
+        return "活跃"
+    if days <= 45:
+        return "留意"
+    if days <= 120:
+        return "陈旧"
+    return "僵尸"
+
+
+GRADE_EMOJI = {"活跃": "🟢", "留意": "🟡", "陈旧": "🟠", "僵尸": "🔴", "未知": "⚪"}
+
+
+def _age_days(item: dict[str, Any]) -> int | None:
+    """优先用 updated_at/created_at 的相对时间估算停滞天数。"""
+    for key in ("updated_at", "created_at", "time_ago", "pr_time"):
+        v = item.get(key)
+        d = parse_relative_days(v) if v else None
+        if d is not None:
+            return d
+    return None
+
+
+def analyze_stale(issues: list[dict[str, Any]], pulls: list[dict[str, Any]]) -> dict[str, Any]:
+    """分析开放 Issue/PR 的陈旧度。"""
+    def _scan(items, kind):
+        out = []
+        for it in items:
+            # 只看开放项
+            if kind == "pr" and it.get("pull_request_status") not in (0, None):
+                continue
+            if kind == "issue":
+                status = str(it.get("issue_status") or "")
+                if "关" in status or "closed" in status.lower():
+                    continue
+            days = _age_days(it)
+            g = grade(days)
+            out.append({
+                "kind": kind,
+                "id": str(it.get("pull_request_number") or it.get("id") or ""),
+                "title": str(it.get("name") or it.get("subject") or it.get("title") or "")[:60],
+                "age_days": days,
+                "grade": g,
+                "author": it.get("author_login") or it.get("author_name") or "",
+            })
+        return out
+
+    records = _scan(issues, "issue") + _scan(pulls, "pr")
+    by_grade: dict[str, int] = {}
+    for r in records:
+        by_grade[r["grade"]] = by_grade.get(r["grade"], 0) + 1
+    # 需要清理的：陈旧 + 僵尸，按停滞天数降序
+    cleanup = sorted(
+        [r for r in records if r["grade"] in ("陈旧", "僵尸")],
+        key=lambda x: x["age_days"] or 0, reverse=True,
+    )
+    return {
+        "total_open": len(records),
+        "by_grade": by_grade,
+        "cleanup_count": len(cleanup),
+        "cleanup": cleanup,
+        "records": records,
+    }
+
+
+def render_report(result: dict[str, Any], owner: str, repo: str) -> str:
+    lines = [
+        f"# 陈旧 Issue/PR 清理报告 — {owner}/{repo}",
+        "",
+        f"开放项共 {result['total_open']} 个。陈旧度分布：",
+        "",
+    ]
+    for g in ("活跃", "留意", "陈旧", "僵尸", "未知"):
+        n = result["by_grade"].get(g, 0)
+        if n:
+            lines.append(f"- {GRADE_EMOJI[g]} {g}：{n}")
+    lines.append("")
+    lines.append("> 分级标准：活跃 ≤14 天 / 留意 ≤45 天 / 陈旧 ≤120 天 / 僵尸 >120 天（按最近更新估算）。")
+    lines.append("")
+    if result["cleanup"]:
+        lines += [f"## 建议处理（{result['cleanup_count']} 个陈旧/僵尸项）", "",
+                  "| 类型 | 编号 | 标题 | 停滞 | 等级 |", "|:----:|:----:|------|:----:|:----:|"]
+        for c in result["cleanup"][:30]:
+            kind = "Issue" if c["kind"] == "issue" else "PR"
+            age = f"{c['age_days']}天" if c["age_days"] is not None else "未知"
+            lines.append(f"| {kind} | {c['id']} | {c['title']} | {age} | {GRADE_EMOJI[c['grade']]}{c['grade']} |")
+        lines += ["", "## 建议行动", "",
+                  "1. 对「僵尸」项：评估是否仍有意义，无意义可关闭或加 `wontfix` 标签。",
+                  "2. 对「陈旧」项：@相关人确认进展，或补充信息后重新激活。",
+                  "3. 关闭操作：`gitlink-cli issue +close --number <web序号>`（写操作，需确认）。", ""]
+    else:
+        lines += ["🎉 没有发现明显陈旧的开放项，积压控制良好。", ""]
+    lines.append("---\n\n由 gitlink-stale 生成。分析只读，关闭操作由维护者确认执行。")
+    return "\n".join(lines)
+
+
+def analyze(owner: str, repo: str, client: GitLinkClient | None = None) -> dict[str, Any]:
+    client = client or GitLinkClient()
+    issues = client.issues(owner, repo, limit=50)
+    pulls = client.pulls(owner, repo, limit=50)
+    return analyze_stale(issues, pulls)
+
+
+def main(argv: list[str] | None = None) -> int:
+    p = argparse.ArgumentParser(prog="gitlink-stale", description="陈旧 Issue/PR 清理")
+    p.add_argument("--owner"); p.add_argument("--repo"); p.add_argument("--slug")
+    p.add_argument("--format", choices=["markdown", "json"], default="markdown")
+    p.add_argument("--output", type=Path)
+    args = p.parse_args(argv)
+
+    if args.slug:
+        owner, repo = split_owner_repo(args.slug)
+    elif args.owner and args.repo:
+        owner, repo = args.owner, args.repo
+    else:
+        print("错误：请用 --owner/--repo 或 --slug 指定仓库。", file=sys.stderr)
+        return 2
+
+    try:
+        result = analyze(owner, repo)
+    except GitLinkError as exc:
+        print(f"采集失败：{exc}", file=sys.stderr)
+        return 1
+
+    out = (json.dumps(result, ensure_ascii=False, indent=2) if args.format == "json"
+           else render_report(result, owner, repo))
+    if args.output:
+        args.output.parent.mkdir(parents=True, exist_ok=True)
+        args.output.write_text(out, encoding="utf-8")
+        print(f"已写入 {args.output}")
+    else:
+        print(out)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/skills/gitlink-stale/tests/test_stale.py b/skills/gitlink-stale/tests/test_stale.py
new file mode 100644
index 0000000..10fa5dd
--- /dev/null
+++ b/skills/gitlink-stale/tests/test_stale.py
@@ -0,0 +1,86 @@
+"""gitlink-stale 单元测试。"""
+from __future__ import annotations
+import sys
+from pathlib import Path
+sys.path.insert(0, str(Path(__file__).resolve().parent.parent / "scripts"))
+import pytest
+from stale import parse_relative_days, grade, analyze_stale, render_report
+
+
+class TestParseRelativeDays:
+    def test_days(self):
+        assert parse_relative_days("3天前") == 3
+
+    def test_weeks(self):
+        assert parse_relative_days("2周前") == 14
+
+    def test_months(self):
+        assert parse_relative_days("2个月前") == 60
+
+    def test_years(self):
+        assert parse_relative_days("1年前") == 365
+
+    def test_recent(self):
+        assert parse_relative_days("刚刚") == 0
+        assert parse_relative_days("3小时前") == 0
+
+    def test_unknown(self):
+        assert parse_relative_days("") is None
+        assert parse_relative_days("奇怪的值") is None
+
+
+class TestGrade:
+    def test_levels(self):
+        assert grade(5) == "活跃"
+        assert grade(30) == "留意"
+        assert grade(90) == "陈旧"
+        assert grade(200) == "僵尸"
+        assert grade(None) == "未知"
+
+
+class TestAnalyzeStale:
+    def test_filters_closed(self):
+        issues = [
+            {"id": "1", "name": "open old", "updated_at": "200天前", "issue_status": "正在解决"},
+            {"id": "2", "name": "closed", "updated_at": "300天前", "issue_status": "关闭"},
+        ]
+        r = analyze_stale(issues, [])
+        # 关闭的不计入
+        assert r["total_open"] == 1
+        assert r["by_grade"]["僵尸"] == 1
+
+    def test_pr_open_only(self):
+        pulls = [
+            {"pull_request_number": "1", "name": "open pr", "pr_time": "100天前", "pull_request_status": 0},
+            {"pull_request_number": "2", "name": "merged", "pr_time": "100天前", "pull_request_status": 1},
+        ]
+        r = analyze_stale([], pulls)
+        assert r["total_open"] == 1
+
+    def test_cleanup_list(self):
+        issues = [{"id": "1", "name": "zombie", "updated_at": "200天前", "issue_status": "open"}]
+        r = analyze_stale(issues, [])
+        assert r["cleanup_count"] == 1
+        assert r["cleanup"][0]["grade"] == "僵尸"
+
+    def test_empty(self):
+        r = analyze_stale([], [])
+        assert r["total_open"] == 0 and r["cleanup_count"] == 0
+
+
+class TestRender:
+    def test_report(self):
+        issues = [{"id": "1", "name": "old issue", "updated_at": "150天前", "issue_status": "open"}]
+        r = analyze_stale(issues, [])
+        md = render_report(r, "o", "r")
+        assert "陈旧 Issue/PR 清理报告" in md
+        assert "建议处理" in md
+
+    def test_clean_repo(self):
+        r = analyze_stale([{"id": "1", "name": "x", "updated_at": "1天前", "issue_status": "open"}], [])
+        md = render_report(r, "o", "r")
+        assert "积压控制良好" in md
+
+
+if __name__ == "__main__":
+    sys.exit(pytest.main([__file__, "-v"]))