AI-powered regression detection for every PR. Analyzes code changes against known bug patterns and flags potential regressions before they ship.
Add to .github/workflows/regressor.yml:
name: Regressor
on:
pull_request:
types: [opened, synchronize, reopened]
jobs:
regressor:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- uses: pemystack/regressor-action@v1
with:
github_token: ${{ secrets.GITHUB_TOKEN }}
# Provide ONE of these:
anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
# openai_api_key: ${{ secrets.OPENAI_API_KEY }}
# gemini_api_key: ${{ secrets.GEMINI_API_KEY }}- Reads the PR diff
- AI analyzes changed code for regression risk patterns
- Checks against known bug patterns (null safety, auth bypasses, concurrency issues, etc.)
- Posts a risk assessment as a PR comment
Works with any language — risk analysis is based on code change patterns, not language-specific tooling.
- 20 reviews/month
- Basic risk analysis
- Upgrade to Pro for unlimited reviews, full analysis, and Slack alerts
| Input | Required | Description |
|---|---|---|
github_token |
Yes | GitHub token for PR comments |
anthropic_api_key |
One of three | Anthropic (Claude) API key |
openai_api_key |
One of three | OpenAI (GPT) API key |
gemini_api_key |
One of three | Google (Gemini) API key |
MIT