Skip to content
Ciying AI Directory
HomeAI toolsGuidesAI newsSkills directoryChatGPT skillsClaude Code skillsCodex skillsTrending in 24 hoursTop 100 skillsDaily updates
Ciying AI Directory
AI DirectorySkills
中文
Ciying AI Directory
HomeAI toolsGuidesAI newsSkills directoryChatGPT skillsClaude Code skillsCodex skillsTrending in 24 hoursTop 100 skillsDaily updates
Skills directory/QwenLM/qwen-code/autofix

Qwen Code Autofix Skill for Local Regression Tests

Review and repair staged, unstaged, and untracked changes within a local Git working tree. The direct skill invocation repeats review and focused checks as needed.

中文

Setup for Qwen Code

Use Qwen Code with its model and sandbox configured, a Git repository, and a POSIX shell. The upstream local workflow requires Git Bash or MSYS on Windows. Use a disposable project with no private content for the example.

  1. Install Qwen Code with `npm install -g @qwen-code/qwen-code@latest` and configure its normal sandbox and model access.
  2. Open a scratch Git project in Qwen Code and select the local /autofix skill. Use the pinned .qwen/skills/autofix directory as the project skill source if the chosen build does not ship it.

Example task

In a toy project with a passing test for empty cart totals, change the empty branch to return undefined. Run /autofix with no arguments from Qwen Code. Review the proposed code scope and proceed only with the scratch files you intend the reviewer to see.

Expected result

A completing review restores the zero total, verifies the focused test, and leaves the corrected files in the local working tree.

Upstream documentation

Original SKILL.md on GitHub · Apache-2.0 license · Source reviewed 2026-11-15

Frequently asked question

Can the Qwen Code Autofix workflow be run by copying it into Codex?

It is documented for Qwen Code and uses that runtime's review command and sandbox. Use Qwen Code for the local /autofix workflow.

Skills directoryTrending 24hTop 100Daily updates
Ciying AI DirectoryAI toolsGuidesAI newsSkills directory