hleliofficiel/ExaAiAgent ? reverse-engineered prompt

Build me a Python security testing app that acts like an AI penetration tester for web apps and APIs.

I want a command line tool that takes a target URL, uses an LLM provider I configure, and runs automated recon, vulnerability checks, proof of concept validation, and a final report. It should use a Docker sandbox for anything risky, keep the host machine safe, and save each run in its own folder with the chat log, run metadata, agent info, and tool activity.

Please make it support multiple agents working together, with sensible defaults so it can auto detect what kind of tests to run. Include a clean interactive terminal experience, clear progress and errors, and output that feels useful to developers and security people. If you need to look up current docs for any libraries or provider setup, do that. Also add tests and a simple install flow so it is easy to run from the command line.