I built an agentic auditor that writes its own Foundry PoCs to prove exploits. Looking for 5 repos to stress test.

Hey everyone,

Like most of you, I’m tired of "AI Auditors" that are just GPT-4 wrappers flagging false positives and non-existent reentrancy.

I’ve been working on a project called Sentinel-1. It’s a multi-agent system (using Claude 3.5/4.6 + an agentic loop) that doesn't just "guess" at vulnerabilities. When it finds a potential logic flaw, it actually spins up a local Foundry environment, writes a .t.sol exploit script, and tries to drain the contract.

If the test passes, the bug is real. If the test fails, it self-corrects the code or discards the finding.

The Goal: I need real-world code to see where the agentic loop breaks.

What you get: A detailed security report + the executable Foundry PoC for any Critical/High bugs found.

How to join: Drop a link to your (public/testnet) GitHub repo below or DM me. I have bandwidth for 5 deep scans this week.

Note: This is focused on Logic Errors, Access Control, and Invariant violations (stuff Slither usually misses).