- BREAKING: Rebranded from Qwen Code to CodeDuet
- BREAKING: Removed
qwenandqwen-clicommands to avoid conflicts with official Qwen CLI - BREAKING: Changed package name to
@anthropic-ai/codeduet-code - Added new primary commands:
codeduet,cd-cli, andcodeduet-code - Enhanced YOLO mode with
--dangerously-skip-permissionsalias - Added environment variable support:
QWEN_YOLO=1 - Added config file setting:
dangerouslySkipPermissions: true - Improved safety warnings for YOLO mode
- Updated documentation and branding throughout
- NEW: Local model support via Ollama and LM Studio
- NEW: Remote GPU support via RunPod.io
- NEW: Complete provider setup guide (PROVIDER_SETUP.md)
- Multi-provider support focus (OpenAI, Anthropic, local models, RunPod, etc.)
- Synced upstream
gemini-clito v0.1.19. - Updated documentation branding from Gemini CLI to Qwen Code.
- Added multilingual docs links in
README.md. - Added deterministic cache control for the DashScope provider.
- Added option to choose a project-level or global save location.
- Limited
grepresults to 25 items by default. grepnow respects.geminiignore.- Miscellaneous improvements and bug fixes.
- Synced upstream
gemini-clito v0.1.18. - Fixed MCP tools.
- Fixed Web Fetch tool.
- Fixed Web Search tool by switching from Google/Gemini to the Tavily API.
- Made tool calls tolerant of invalid-JSON parameters occasionally returned by the LLM.
- Prevented concurrent query submissions in rare cases.
- Corrected Qwen logger exit-handler setup.
- Separated static QR code and dynamic spinner components.
- Added usage statistics logging for Qwen integration.
- Made
/initrespect the configured context filename and aligned docs withQWEN.md. - Fixed
EPERMerror when runningqwen --sandboxon macOS. - Fixed terminal flicker while waiting for login.
- Fixed
glm-4.5model request error.
- Added Qwen OAuth login and up to 2,000 free requests per day.
- Synced upstream
gemini-clito v0.1.17. - Added the
systemPromptMappingsconfiguration option.