feat: add Qwen Code support to Spec Kit

[ahmet-cetinkaya]

Summary

• Add comprehensive Qwen Code integration as fourth AI assistant option
• Extend CLI, documentation, and build pipeline to support Qwen
• Maintain backward compatibility with existing assistants (Claude, Gemini, Copilot)

Changes Made

• CLI Integration: Added Qwen as option in src/specify_cli/__init__.py with proper help text and tool checking
• Shell Scripts: Extended scripts/update-agent-context.sh to support QWEN.md generation
• GitHub Actions: Added Qwen package creation in release workflow with .qwen/commands/*.md structure
• Documentation: Updated README.md, CONTRIBUTING.md, and templates to include Qwen CLI references
• Templates: Extended plan template to reference Qwen in agent file updates

Testing

• ✅ CLI help shows Qwen option: --ai claude, gemini, copilot, or qwen
• ✅ All commands function correctly with Qwen integration
• ✅ Check command runs without errors
• ✅ Follows established patterns from other AI assistants

Implementation Details

• Command Structure: Uses .qwen/commands/*.md format
• Context File: Generates QWEN.md in project root
• CLI Tool: Expects qwen command availability
• Installation URL: References https://github.com/QwenLM/qwen-code

This extends Spec Kit's AI assistant ecosystem to include Qwen while maintaining full compatibility with existing workflows.

[nds15763]

Truly appreciated

[vjaykrsna]

Can't wait to get qwen code support as well

[vjaykrsna]

@ahmet-cetinkaya is this complete?
because i tried uvx --from git+https://github.com/ahmet-cetinkaya/spec-kit#feature/add-qwen-support specify
but i didn't see qwen code cli there
guide me if i did it wrong

[ahmet-cetinkaya]

@vjaykrsna There’s a small mistake in your command — when specifying a branch, you should use @ instead of #. Try this:

uvx --from git+https://github.com/ahmet-cetinkaya/spec-kit@feature/add-qwen-support specify

This will let you see the new option.

However, since spec-kit fetches its templates from releases, the tool won’t fully work this way.
To properly run the feature before release, you’ll need to clone the branch locally:

git clone https://github.com/ahmet-cetinkaya/spec-kit -b feature/add-qwen-support
cd spec-kit

uv sync
uv run specify

[vjaykrsna]

by default it fetch from github had to modify the function to get it to work after generating the files manually
hope they officially merge this soon

[localden]

@ahmet-cetinkaya once you address the nit changes I highlighted, I am happy to merge this.

[ahmet-cetinkaya]

Hi @localden,

The requested changes are applied, and the branch is now ready to merge. Appreciate the review! 🚀

I usually keep lists numbered sequentially so they’re easier to read in raw Markdown, but if the 1. style is preferred, I’m happy to stick with that.

[localden]

@ahmet-cetinkaya - last request, can you rev the CLI version and update the changelog with the addition of Qwen support?

[ahmet-cetinkaya]

@localden I've addressed your last request:

1. Bumped the CLI version from 0.0.4 to 0.0.5 in pyproject.toml
2. Updated the CHANGELOG.md with the new version and added Qwen Code support as a new feature

The changes have been pushed to the branch. This should be ready to merge now.

[vjaykrsna]

Waiting for the PR to merge feels like forever.
some good ones are find for quite some time now

[localden]

Thank you for contribution, @ahmet-cetinkaya! 🥳

[aemr3]

hey @ahmet-cetinkaya, @localden there are some missing definitions to support qwen, please check my pr: #64

Important ones are update-agent-context scripts, create-release-packages.sh, release.yml as far as I can see.

[SeanPtest]

I tried this moments ago, didn't work.

uvx --from git+https://github.com/github/spec-kit.git specify init spec-sample
Selected AI assistant: qwen
Selected script type: sh
No matching release asset found for pattern: spec-kit-template-qwen-sh
... ... ... spec kit template-qwen.sh.zip isn't listed, the others are.
Initialization failed: 1

I did a search for the missing zip file in the merged repo, didn't find it. We can uhh select qwen tho, at least.

Tried it like this also, same error.

uvx --from git+https://github.com/github/spec-kit.git specify init spec-qwen --ai qwen --script sh

[ahmet-cetinkaya]

@aemr3 @SeanPtest - Fixed! PR #321 adds missing Qwen assets to releases. specify init --ai qwen will work after merge. Thanks!

[Mingholy]

#330 Changed md to toml so that the FileCommandLoader can load spec commands into qwen-code.