I built a GitHub Action that auto-reviews PRs with LLM — risk assessment + evidence mapping [alpha, OSS]

I built a GitHub Action that analyzes pull requests automatically and posts structured comments.

What it does

• Risk assessment (low/medium/high) based on file patterns and diff analysis
• Maps evidence to specific line numbers in the diff
• Detects security patterns: CVEs, broad exception handling, TLS misconfigurations
• Posts comment automatically on every PR

Example comment it leaves on a PR

🟢 Risk: LOW

What: mypy upgraded from 1.3 to 1.5.1 in pyproject.toml
Why: Address bugs in mypy 1.5.1

Evidence:

• pyproject.toml:L117 — mypy==1.5.1

Install (~2 minutes)

Just add the workflow file and two secrets to your repo. Full instructions in the README.

Repo: https://github.com/mordecaiusm922-create/devmind

Still alpha. Looking for feedback on what's useful and what's noise.

Source: Dev.to