cathaloconnorrh/AI Dependency Fixer
Automatically fix breaking changes from dependency update PRs using AI
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Activelast commit Mar 12, 2026
- License
- None
Pinned Snippet
uses: cathaloconnorrh/ai-dependency-fixer@339374264e09e23d87f97eebdd7ca7463523d1f0 # v1tags can be moved; commit SHAs can't. why a SHA?
Inputs
| name | description | required | default |
|---|---|---|---|
| ai-api-key | API key for the chosen AI provider | yes | — |
| ai-provider | AI provider to use: anthropic, openai, or openai-compatible | no | anthropic |
| ai-base-url | Base URL for OpenAI-compatible self-hosted models (e.g. http://localhost:11434/v1) | no | "" |
| mode | Operating mode: "fix" only fixes failing tests, "investigate" also proactively updates passing code | no | investigate |
| max-attempts | Maximum number of fix attempts before giving up | no | 3 |
| test-command | Test command to run. Use "auto" for automatic detection | no | auto |
| install-command | Dependency install command. Use "auto" for automatic detection | no | auto |
| max-diff-lines | Maximum lines in AI-generated diff before rejecting (safety guardrail) | no | 200 |
| ai-model | Model to use (defaults depend on provider: claude-sonnet-4-20250514 for Anthropic, gpt-4o for OpenAI) | no | "" |
| github-token | GitHub token for PR comments | no | ${{ github.token }} |
Outputs
| name | description |
|---|---|
| result | Result: "fixed", "already-passing", or "failed" |
| attempts | Number of fix attempts made |