Posts

Can we really automate with AI? P2: How I actually built one

I roasted every AI pentester on GitHub in P1, so heres how I actually built mine: real Burp Suite, three MCP servers, machine-verified evidence, and a coverage gate the LLM cant talk its way past.

Can we really automate with AI? P1: System Prompt as a Pentester

Tearing apart every ‘AI pentester’ on GitHub - T3MP3ST, Strix, PentestGPT and friends. Theyre all the same thing: a system prompt saying ‘you are a hacker’. Code-level proof inside.

Can we really automate with AI? P0: Why am I writing this

Kicking off a series on whether AI can actually replace web pentesters - who I am, why I’m writing this, and what’s coming in P1-P3.