Claude AutoFix
Version updated for https://github.com/enriconunes/claude-autofix-action to version v2.1.
- This action is used across all versions by ? repositories.
Action Type
This is a Composite action.
Go to the GitHub Marketplace to find the latest changes.
Action Summary
The Claude AutoFix Action is a GitHub Action that automates the identification and resolution of test failures in pull requests. It uses Claude AI to analyze failing tests, provide diagnostic comments, generate fixes, and create a new pull request with corrected code for review. This streamlines the debugging process and reduces manual effort for developers in projects using Python (pytest), TypeScript, or JavaScript (Vitest/Jest).
What’s Changed
This release introduces error logging and AI-powered developer insights.
What’s new:
- Error log (enable-error-log) — every failing test is recorded as a structured entry in logs/error_history.jsonl on the main branch, capturing timestamp, author, branch, file, test name, error type, error message, and whether the fix was applied. No Claude tokens used.
- Developer insights (enable-insights) — after 3 or more recorded failures, Claude analyses the author’s error history and posts a comment on the PR pointing out recurring patterns and giving concrete, actionable tips. Uses Claude Haiku to keep costs low.
Both features are opt-in and off by default. Existing setups require no changes to upgrade.