I spend a lot of time working with AI coding agents Claude Code, Cline, Cursor, OpenCode. And every time I start a new task in a large repo, I hit the same wall: the agent has no idea what the codebase looks like.
So I built Sentinel.
What it does
Point it at any repo. It scans everything locally (no uploads, no API keys, no internet needed) and produces:
- 🏗️ Architecture map — components, dependencies, patterns
- 💊 Health score — maintainability, complexity, test coverage
- 🔥 Risk hotspots — oversized files, TODO density, doc drift
- 🎯 Entry points — what the AI should focus on first
- 🤖 Agent prompt — ready-to-paste into Claude Code / Cline / Cursor
- 📄 HTML report — self-contained, zero external assets
Why local-only matters
Every scan runs entirely on your machine. No code leaves your disk. No API calls. Pure Python stdlib — zero external dependencies.
The AI agent prompt it generates a total of (~2,500 tokens) replaces hours of manually reading files and explaining the codebase to your agent.
Get it
bash
pip install git+https://github.com/Ntooxx/Sentinel.git
project-sentinel scan . --fast
🔗 GitHub: https://github.com/Ntooxx/Sentinel
🌐 Dashboard demo: sentinel-nt.netlify.app
📦 197 tests, 0 failures
Let me know what breaks or what's missing. I'm actively improving it.