Commit graph

2500 commits

Author SHA1 Message Date
tanzhenxin
cba9c424eb fix(core): handle missing delta in OpenAI stream chunks
Some OpenAI-compatible providers occasionally emit chat.completion.chunk choices
without a delta object. Guard optional reasoning_content access and add a
regression test to ensure chunk conversion does not throw.
2026-01-09 17:41:07 +08:00
tanzhenxin
6714f9ce3c
Merge pull request #1351 from xuewenjie123/fix/editor-launch-issues
fix: resolve external editor launch failure on macOS and Windows
2026-01-09 11:07:43 +08:00
tanzhenxin
155d1f9518
Merge pull request #1428 from liqiongyu/fix/727-memory-show-respects-context-file
fix(cli): /memory show respects context.fileName
2026-01-09 10:29:27 +08:00
mingholy.lmh
36c142951a fix: default authType fallback 2026-01-08 18:38:23 +08:00
mingholy.lmh
2b511d0b83 fix: cold start issue and acp integration tests 2026-01-08 18:08:56 +08:00
mingholy.lmh
85bc0833b4 fix: remove authType fallback option for cold start case 2026-01-08 18:08:56 +08:00
tanzhenxin
2662639280
Merge pull request #1424 from xuewenjie123/refactor/ide-context-plain-text
refactor: convert IDE context from JSON to plain text format
2026-01-08 17:11:31 +08:00
Mingholy
b7ac94ecf6
Merge pull request #1291 from QwenLM/mingholy/feat/multi-provider-support
feat: multi-provider models config support
2026-01-08 16:27:00 +08:00
tanzhenxin
be8259b218
Merge pull request #1433 from BlockHand/userandwork
feat: Modify the selection order of user Settings and workspace Settings
2026-01-08 16:07:02 +08:00
xwj02155382
ca4c36f233 feat: wrap selected text in code blocks for IDE context 2026-01-08 16:06:21 +08:00
Mingholy
f41308f34c
Merge pull request #1426 from QwenLM/fix/oauth-headless-auth-url-display
fix(core): ensure OAuth URL always displayed in headless mode
2026-01-08 16:03:05 +08:00
刘伟光
82cbdee3b4 feat: 恢复组件使用修改默认值 2026-01-08 15:35:17 +08:00
mingholy.lmh
81de79c899 fix: best effort to use resolved authType/model across the repo 2026-01-08 12:11:23 +08:00
刘伟光
f6a753cf78 feat: Modify the selection order of user Settings and workspace Settings 2026-01-08 11:50:55 +08:00
刘伟光
509d304742 feat: Modify the selection order of user Settings and workspace Settings 2026-01-08 11:36:07 +08:00
刘伟光
6319a6ed56 feat: 修改用户设置和工作区设置的选择顺序 2026-01-08 10:45:22 +08:00
mingholy.lmh
ab07c2d89c chore: bump version to 0.7.0 2026-01-08 10:21:55 +08:00
mingholy.lmh
5ea841dd02 fix: refine auth message to give explicit tip 2026-01-07 22:58:11 +08:00
mingholy.lmh
ded1ebcdff fix: fallback and auth issues when configuring a duplicate model id 2026-01-07 22:58:11 +08:00
mingholy.lmh
afe6ba255e fix: align authType & model persisting behavior across dialogs 2026-01-07 22:58:11 +08:00
mingholy.lmh
8da376637a fix: remove detailed generationConfig 2026-01-07 22:58:10 +08:00
mingholy.lmh
15f4c1ebd6 chore: add i18n 2026-01-07 22:58:10 +08:00
mingholy.lmh
492da0c8c0 chore: update copyright notice in modelConfigUtils.ts 2026-01-07 22:58:09 +08:00
mingholy.lmh
90855c93d1 fix: lint & ci issues 2026-01-07 22:58:09 +08:00
mingholy.lmh
db12796df5 refactor: update authentication handling and model configuration
- Enhanced authentication method validation in `auth.ts` and `auth.test.ts`.
- Introduced new model provider configuration logic
- Updated environment variable handling for various auth types.
- Removed deprecated utility functions and tests related to fallback mechanisms.
2026-01-07 22:58:09 +08:00
mingholy.lmh
aa9cdf2a3c review: stage1 2026-01-07 22:58:08 +08:00
liqoingyu
0a0ab64da0 test(cli): make memoryCommand path assertions cross-platform 2026-01-07 20:28:28 +08:00
liqoingyu
8a15017593 fix(cli): /memory show respects context.fileName 2026-01-07 20:07:41 +08:00
LaZzyMan
4d54a231b3 fix(core): ensure OAuth URL always displayed in headless mode
- Always show authentication URL before attempting browser launch
- Fixes issue where browser launch silently fails in headless environments
- Improves error logging for browser launch failures

Fixes #1425
2026-01-07 19:37:47 +08:00
tanzhenxin
570ec432af
Merge pull request #1282 from BlockHand/fix-sandbox-ideInstall
feat: Optimize the issue where an error message indicating unfriendli…
2026-01-07 17:40:45 +08:00
tanzhenxin
bfc3bbfa9c update user messages 2026-01-07 17:25:27 +08:00
tanzhenxin
91af9bf6c8 Merge branch 'main' into fix-sandbox-ideInstall 2026-01-07 17:12:22 +08:00
tanzhenxin
f6771c0858
Merge pull request #1393 from Weaxs/main
[OpenaiContentGenerate] convertOpenAIResponseToGemini record thoughtsTokenCount
2026-01-07 17:05:28 +08:00
tanzhenxin
2c8be05029
Merge pull request #1415 from QwenLM/fix/openai-reasoning-config
fix(core): don’t force reasoning/topP defaults for OpenAI-compatible APIs
2026-01-07 16:59:26 +08:00
tanzhenxin
4744af1ea8
Merge pull request #1406 from QwenLM/fix/non-interactive-tool-permission
fix(cli,core): honor `tools.core` / `tools.allowed` in non-interactive runs
2026-01-07 16:59:19 +08:00
Mingholy
2c285394c7
Merge pull request #1423 from QwenLM/chore/release-v0.6.1
chore: bump version to 0.6.1
2026-01-07 16:55:45 +08:00
xwj02155382
0f1cb162c9 refactor: convert IDE context from JSON to plain text format
Fixes #1418

- Remove JSON.stringify() and code fences from getIdeContextParts()
- Use human-readable plain text format for better LLM comprehension
  - Full context: 'Active file:', 'Cursor: line X, character Y'
  - Delta updates: 'Files opened:', 'Files closed:', 'Cursor moved:', etc.
- Update all related tests to match new plain text format
- All 49 tests passing

This change improves the model's ability to read and reason about IDE
state by eliminating escaped characters and rigid JSON structure that
can confuse LLMs when interpreting file paths, cursor positions, or
selection ranges.
2026-01-07 16:46:48 +08:00
xwj02155382
3d059b71de refactor: improve IDE context format and editor command error handling
- Change IDE context from JSON to plain text format for better LLM comprehension
  - Remove JSON.stringify() and code fences from getIdeContextParts()
  - Use human-readable format: 'Active file:', 'Cursor: line X, character Y'
  - Apply same format to delta updates: 'Files opened:', 'Files closed:', etc.
  - Update all related tests to match new plain text format

- Fix editor command fallback logic in useLaunchEditor
  - Throw clear error when no editor command is available
  - Remove meaningless fallback to last command in list
  - Provide helpful error message with tried commands and solution
2026-01-07 16:34:12 +08:00
mingholy.lmh
f2d941e469 chore: bump version to 0.6.1 2026-01-07 16:25:14 +08:00
tanzhenxin
9b2dfe1e06
Merge pull request #1374 from QwenLM/fix/resume-command-broken-after-new-chat
Fix resume command broken after new chat
2026-01-07 16:21:47 +08:00
xwj02155382
177a91f1d5 Merge branch 'main' of github.com:QwenLM/qwen-code into fix/windows-background-terminal-execute-x 2026-01-07 15:50:03 +08:00
Weaxs
870d207f18 revert enable_thinking & thinking_budget 2026-01-07 10:25:34 +08:00
Tu Shaokun
0878ee4cbd fix: handle setApprovalMode error in untrusted folders
Add try/catch to gracefully handle errors when setting privileged
modes (yolo/auto-edit) in untrusted folders, returning an error
message instead of throwing.
2026-01-06 22:04:43 +08:00
Tu Shaokun
bfe7298858 refactor: apply session-only approval mode per review feedback
- Remove persistence to user settings (no setValue call)
- Only use config.setApprovalMode() for session scope
- Remove autocomplete feature for simplicity
- Align with Shift+Tab behavior
2026-01-06 21:57:21 +08:00
Tu Shaokun
2f2937aafe test: add explicit assertions for setApprovalMode argument
Verify the exact mode value passed to config.setApprovalMode to catch
potential regressions in settings merge/update mechanism.
2026-01-06 21:54:33 +08:00
Tu Shaokun
8fcdd86b91 feat(cli): add direct argument support for /approval-mode command
Allow users to set approval mode directly via argument instead of
opening the dialog. For example:
- /approval-mode plan
- /approval-mode yolo
- /approval-mode auto-edit
- /approval-mode default

If no argument is provided, the dialog opens as before.
If an invalid argument is provided, an error message shows valid options.

Also adds tab completion for mode arguments.

Fixes #1353
2026-01-06 21:54:33 +08:00
tanzhenxin
d7d7bf0c39 fix default values of reasoning config for openai compatible api 2026-01-06 19:39:28 +08:00
gwinthis
b95d9a8d2d
Merge pull request #1414 from QwenLM/doc/qwencode-java
Doc/qwencode java
2026-01-06 17:54:47 +08:00
顾盼
6f39ae120c
Merge pull request #1355 from QwenLM/feat/stable-acp-flag
feat: graduate `--experimental-acp` to stable `--acp` flag
2026-01-06 17:51:43 +08:00
顾盼
627857621a
Merge pull request #1365 from QwenLM/fix/missing-whitespaces
fix: preserve whitespace in thinking content for stream-json output format
2026-01-06 17:51:26 +08:00