Aivist/Aivist-Verify ? reverse-engineered prompt

Build me a Python 3.11 command line tool for confirming BOLA and IDOR issues, not just flagging possible ones.

I want it to take a candidate endpoint and two identities, send the real requests, and decide whether one user can actually cross into another user’s data. The important part is that the final verdict must be made by deterministic code, not by the model, and it should never be able to mark something verified unless the evidence really supports it. If the check is inconclusive or blocked, it should say so honestly instead of guessing.

Please include a small demo mode that runs against a built in vulnerable target, plus clear output that shows the request, response, and a reproducible evidence chain the user can rerun later. Make it easy to run locally, with a simple CLI entry point and sensible defaults. If you need to look up current docs or best practices online, go ahead.