694695050a
fix(opencode): preserve tool input from running state for MCP tool results ( #9667 )
2026-01-20 15:12:15 -06:00
68e41a1ee7
fix: pass arguments to commands without explicit placeholders ( #9606 )
2026-01-20 11:12:43 -06:00
5622c53e1f
tweak: adjust codex prompt to discourage unnecessary question asking and encourage more autonomy
2026-01-20 11:08:04 -06:00
419004992d
chore: remove duplicate prompt file
2026-01-19 23:22:57 -06:00
0d49df46ef
fix: ensure truncation handling applies to mcp servers too
2026-01-19 23:19:24 -06:00
bfd2f91d5b
feat(hook): command execute before hook ( #9267 )
2026-01-18 13:11:22 -06:00
b7ad6bd839
feat: apply_patch tool for openai models ( #9127 )
2026-01-18 00:35:09 -06:00
9127055ae7
tweak: wording
2026-01-16 15:50:24 -06:00
f5a6a4af7f
Revert "fix: ensure that tool attachments arent sent as user messages ( #8944 )"
...
This reverts commit 8fd1b92e6e
2026-01-16 15:50:24 -06:00
14d1e20287
Revert "fix(app): support anthropic models on azure cognitive services" ( #8966 )
2026-01-16 15:25:23 -06:00
b8e2895dfc
fix(app): support anthropic models on azure cognitive services ( #8335 )
2026-01-16 15:24:06 -06:00
8fd1b92e6e
fix: ensure that tool attachments arent sent as user messages ( #8944 )
2026-01-16 12:47:43 -06:00
e0a854f035
Revert "fix: rm user message when dealing w/ image attachments, use proper tool attachment instead"
...
This reverts commit de2de099b4
2026-01-16 09:33:54 -06:00
524ea95a00
update gpt models prompt
2026-01-16 01:34:01 -06:00
de2de099b4
fix: rm user message when dealing w/ image attachments, use proper tool attachment instead
2026-01-16 00:05:10 -06:00
d7192d6af9
tweak: set opencode as user agent for most interefence requests
2026-01-15 19:25:58 -06:00
8b08d340ac
fix: stop changing main model/agent from subtasks invocation ( #7681 )
...
Co-authored-by: Aiden Cline <aidenpcline@gmail.com>
2026-01-15 17:18:39 -06:00
9b57db30d1
feat: add litellmProxy provider option for explicit LiteLLM compatibility ( #8658 )
...
Co-authored-by: Mark Henderson <Mark.Henderson99@hotmail.com>
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
2026-01-15 22:01:15 +00:00
b14622352e
fix(session): ensure agent exists before processing title in session summary ( #8662 )
2026-01-15 15:24:13 -06:00
8d720f9463
fix(opencode): add input limit for compaction ( #8465 )
2026-01-15 01:35:16 -06:00
92931437c4
fix: codex id issue ( #8605 )
2026-01-15 01:31:50 -06:00
6e13e2f74e
fix(session): remove typo'd duplicate path import ( #8408 ) ( #8412 )
...
Co-authored-by: opencode <opencode@sst.dev>
Co-authored-by: Adam <2363879+adamdotdevin@users.noreply.github.com>
2026-01-14 09:04:03 -06:00
fcc561ebb7
fix plan mode when not in git worktree
2026-01-14 08:21:26 -05:00
f9fcdead55
fix(session): skip duplicate system prompt for Codex OAuth sessions ( #8357 )
...
Co-authored-by: Claude <noreply@anthropic.com>
2026-01-13 23:44:39 -06:00
0a3c72d678
feat: add plan mode with enter/exit tools ( #8281 )
2026-01-13 15:55:48 -05:00
d954e1e3b6
fix(session): store api url in metadata ( #8090 )
2026-01-12 17:22:42 -06:00
71a7ad1a4e
core: fix model selection in title generation to use user's model instead of assistant's
2026-01-12 12:45:24 -05:00
8b9a85b7e7
fix(mcp): support `resource` content type in MCP tool output ( #7879 )
2026-01-11 21:39:42 -06:00
4752c83155
feat: pass sessionID to chat.system.transform ( #7718 )
2026-01-10 18:21:51 -06:00
8b287caaad
tweak codex instructions prompt for codex plan
2026-01-09 18:07:24 -06:00
172bbdaced
feat: codex auth support ( #7537 )
2026-01-09 17:47:37 -06:00
a5edf3a311
fix: resolve broken forked sessions with compactions due to missing parent-child message references ( #6445 )
2026-01-09 11:42:55 -06:00
542c9d5346
feat(flags): allow disabling .claude prompt and skills loading ( #7205 )
2026-01-08 00:10:25 -06:00
e37fd9c105
core: add interactive question tool for gathering user preferences and clarifying instructions ( #7268 )
2026-01-07 22:29:42 -05:00
2e4fe973c9
fix: issue w/ normal transform options conflicting w/ small model options when gen-ing title
2026-01-07 17:32:38 -06:00
1b82511fbd
feat: write truncated tool outputs to files ( #7239 )
2026-01-07 17:25:00 -06:00
34c9d106ee
refactor: simplify task tool subagent filtering ( #7165 )
...
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
2026-01-07 13:28:13 -06:00
429240f439
ignore: add truncation funcs ( #7178 )
2026-01-07 01:43:06 -06:00
fd7b7eacd3
Added: Ability to hide subagents from primary agents system prompt. ( #4773 )
...
Co-authored-by: GitHub Action <action@github.com>
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
Co-authored-by: Aiden Cline <aidenpcline@gmail.com>
2026-01-06 22:29:17 -06:00
625c9dae5c
fix: ensure commands listing subagent work
2026-01-06 17:05:20 -06:00
32e0b612d9
adding timeout ( #7128 )
2026-01-06 15:00:34 -06:00
a10cc63403
feat: url based instructions ( #7125 )
2026-01-06 14:48:22 -06:00
630476afc0
load `OPENCODE_CONFIG_DIR` AGENTS.md into the system prompt ( #7115 )
2026-01-06 13:43:55 -06:00
5db78f20e9
core: fix title generation for subtask-only messages to extract actual user prompts instead of generic tool execution descriptions
2026-01-06 13:30:39 -06:00
485aadcbfa
fix: restore skill filtering by agent permissions ( #7042 )
2026-01-06 12:01:37 -06:00
a38e1701ee
fix: pass image parts to custom commands ( #6525 )
...
Co-authored-by: Melih Mucuk <melih@monkeysteam.com>
2026-01-05 13:06:57 -06:00
554572bc39
fix: prevent main model thinking variant from applying to small model ( #6839 )
...
Co-authored-by: Melih Mucuk <melih@monkeysteam.com>
2026-01-04 13:28:22 -06:00
21dc3c24d9
feat: mcp resources ( #6542 )
2026-01-04 09:12:54 -06:00
5aa4fd0042
core: add variant to chat.message input
2026-01-04 00:28:52 -06:00
f991fbbde8
core: ephemerally wrap queued user messages with reminder to stay on track ( #6725 )
2026-01-02 22:42:56 -06:00
3c41e4e8f1
chore: rename repo references from sst/opencode to anomalyco/opencode ( #6687 )
...
Co-authored-by: Github Action <action@github.com>
2026-01-02 16:02:52 -05:00
c996f3d847
chore: ensure empty message isnt sent
2026-01-02 12:28:40 -06:00
b84a1f714b
tui: remove memory leak fixes documentation after implementation
2026-01-01 21:40:28 -05:00
07c008fe3d
chore: generate
2026-01-02 02:28:48 +00:00
dad9c917d2
tui: fix memory leaks in session management and improve permission error handling
2026-01-01 21:28:11 -05:00
351ddeed91
Permission rework ( #6319 )
...
Co-authored-by: Github Action <action@github.com>
Co-authored-by: Adam <2363879+adamdotdevin@users.noreply.github.com>
2026-01-01 17:54:11 -05:00
7a3ff5b98f
fix(session): check for context overflow mid-turn in finish-step ( #6480 )
2026-01-01 12:03:18 -06:00
3b03324578
fix: display error if invalid agent is used in a command ( #6578 )
2026-01-01 11:39:21 -06:00
9a1dc1ffe4
core: prevent TimeoutOverflowWarning by capping setTimeout delay to max 32-bit signed integer
2025-12-31 16:15:49 -06:00
57a2b5f444
feat: mcp prompts as slash commands (alternative) ( #5767 )
...
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
Co-authored-by: Aiden Cline <aidenpcline@gmail.com>
2025-12-31 00:51:25 -06:00
81fef60266
fix: ensure variants also work for completely custom models ( #6481 )
...
Co-authored-by: Daniel Smolsky <dannysmo@gmail.com>
2025-12-30 16:37:32 -06:00
8f629db988
feat: add extract reasoning middleware ( #6463 )
2025-12-30 13:13:18 -06:00
8b708242f1
chore: fix various typos ( #6429 )
2025-12-30 09:42:04 -06:00
ed0c0d90be
feat: add variants toggle ( #6325 )
...
Co-authored-by: Github Action <action@github.com>
2025-12-29 21:43:50 -06:00
b157fd10a7
fix: filter messages with only step-start parts in toModelMessage ( #6383 )
2025-12-29 14:58:11 -06:00
29d8557d41
doc: fix typos in various files ( #6294 )
2025-12-28 13:06:25 -06:00
79c263494f
tweak: inform agent if no skills are available
2025-12-27 21:20:00 -06:00
e6b9988fa4
doc: fix typos in various files ( #6238 )
2025-12-27 00:46:06 -06:00
2b054bec95
core: fix compaction config checks to properly respect user settings
2025-12-26 19:48:56 -05:00
ed06de5e30
core: add configurable compaction settings to allow users to disable auto-compaction and pruning via config instead of flags
2025-12-26 19:31:48 -05:00
25c68c8061
chore: kill the dead Polaris Alpha code ( #6193 )
2025-12-26 11:32:31 -06:00
cf38884778
doc: fix typos in various files ( #6196 )
2025-12-26 10:21:33 -06:00
99e2112807
tweak: retry err
2025-12-23 22:10:28 -06:00
7396d495ee
chore: regen sdk
2025-12-23 19:34:38 -06:00
12ee9d51c3
make 'The socket connection was closed unexpectedly' errors retryable
2025-12-23 19:20:31 -06:00
eab177f5e7
feat(plugin): allow compaction hook to replace prompt entirely ( #5907 )
2025-12-22 22:19:14 -06:00
3a54ab68d1
feat(skill): add per-agent filtering to skill tool description ( #6000 )
2025-12-22 20:14:33 -05:00
046e351140
feat: add native skill tool with permission system ( #5930 )
...
Co-authored-by: Dax Raad <d@ironbay.co>
2025-12-22 18:24:06 -05:00
d4b7f75ce3
fix: Perform snapshot in cases finish-step is not reached ( #5912 )
...
Co-authored-by: noamzbr <noamzbr@users.noreply.github.com>
2025-12-21 21:13:11 -06:00
8fe0715928
feat: add Agent Skills support ( #5921 )
2025-12-21 19:44:56 -05:00
c333ffa38b
core: fix LSP server binary installation and shell command execution
...
- Ensure proper file permissions are set for installed LSP binaries on non-Windows platforms
- Add error handling for shell command execution in prompt system to prevent crashes
2025-12-21 18:33:37 -05:00
8f6c8844d7
feat: support configuring a default_agent across all API/user surfaces ( #5843 )
...
Co-authored-by: observerw <observerw@users.noreply.github.com>
2025-12-20 11:49:23 -06:00
a7a2bbb497
feat: add endpoints to delete and update message parts ( #5433 )
2025-12-20 15:00:41 +00:00
4fd576f3af
fix: better api call error msgs in some cases
2025-12-18 18:46:25 -06:00
ecc5050838
tweak: more retry cases
2025-12-18 13:59:37 -06:00
4bf882ba81
fix(command): validate model before executing slash command ( #5740 )
2025-12-18 10:35:40 -06:00
24430287c5
feat(plugin): add experimental.session.compacting hook for pre-compaction context injection ( #5698 )
...
Co-authored-by: Aiden Cline <aidenpcline@gmail.com>
2025-12-17 13:57:09 -06:00
69e562125d
fix(tui): resolve session_status TypeError ( #5520 )
2025-12-17 11:38:05 -06:00
1e4bfbcf6f
add OPENCODE_EXPERIMENTAL_OUTPUT_TOKEN_MAX to override 32k default ( #5679 )
...
Co-authored-by: qio <handsomehust@gmail.com>
2025-12-17 10:35:43 -06:00
b8204c0bb7
fix: config option setCacheKey not being respected ( #5686 )
2025-12-17 10:20:10 -06:00
b8a8fb0de6
plugin(hook): add task tool execution hooks and command context tracking ( #5642 )
2025-12-16 19:28:09 -06:00
5f57cee8e4
fix: user invoked subtasks causing tool_use or missing thinking signa… ( #5650 )
2025-12-16 15:42:21 -06:00
c2944024a8
fix: enable shell alias expansion in ! command ( #5621 )
2025-12-16 11:32:31 -06:00
11b3927dc2
fix: use system prompt field from prompt input ( #5633 )
...
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
2025-12-16 10:52:22 -06:00
72ebaeb8f7
fix: rejoin system prompt if experimental plugin hook triggers to preserve caching ( #5550 )
...
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
2025-12-15 20:00:26 -06:00
81134cf61e
add ability to set topK
2025-12-15 14:34:56 -06:00
b021b26e77
feat: restore experimental.chat.messages.transform and add experimental.chat.system.transform hooks ( #5542 )
2025-12-14 22:51:11 -06:00
54569b5552
fix(session): fix unshare command not clearing share state ( #5523 )
2025-12-14 22:05:06 -06:00
fed4776451
LLM cleanup ( #5462 )
...
Co-authored-by: GitHub Action <action@github.com>
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
2025-12-14 21:11:30 -05:00
7368342bab
feat: add experimental.continue_loop_on_deny config option ( #4729 )
...
Co-authored-by: Aiden Cline <aidenpcline@gmail.com>
2025-12-14 17:13:32 -06:00
Trevor Walker