unclecheng-li/vulnclaw — reverse-engineered prompt

Build me a Python based AI penetration testing assistant for authorized targets only. I want to be able to type something simple like “test this site” and have it handle the whole flow on its own, from basic recon and fingerprinting, to finding likely vulnerabilities, trying safe proof of concept exploitation, and then generating a clean report with evidence and a runnable Python PoC when relevant.

Please make it work as both a CLI and a simple local web app. I also want a terminal workspace, session history, saved targets, configuration for different OpenAI compatible model providers and API keys, and a doctor check that tells me if my environment is ready. It should support optional browser automation and traffic replay tools, but still be useful out of the box.

The important part is that it should be goal driven, not just loop forever, and it should only claim success if the result is clearly backed by real tool output. Add plugin style vulnerability checks, useful encode and crypto helpers, report generation, and Docker support. Look up current docs online if you need to.