Multi-Agent AI Security Bot for GitHub Repositories
Patchy automates vulnerability detection, generates precise code fixes, and manages security issues in your GitHub repos using a multi-agent AI architecture powered by Cerebras LLM.
- 🔍 Auto-scan vulnerabilities — Run Semgrep security scans on your repositories
- 🔧 Generate fixes automatically — AI-powered code patches for detected issues
- 📋 Create PRs — Automatic pull requests with fixes ready to merge
- 💬 Issue interaction — @mention Patchy in GitHub issues for AI-powered analysis
## Tech Stack
| Layer | Technology |
|---|---|
| Backend | Flask (Python) |
| Frontend | HTML + TailwindCSS + Vanilla JS |
| Auth | GitHub OAuth |
| Scanner | Semgrep |
| LLM | Cerebras (qwen-3-235b) |
| Storage | JSON file |
Patchy extends the multi-agent security architecture from SentryAgent (smart contract security) to general application security, adding:
- Web interface with GitHub integration
- Broader language support
- Automated PR creation
- Conversational issue interaction
MIT