Testing AI agents where they break.

I'm Matheus, a AI security engineer working at Avenza Cloud, building RedThread on the side—an autonomous red-teaming engine for testing LLM systems.

Writing

Writing about my thoughts and experiences using AI, both in enterprise and greenfield projects:

• How I used Claude Code to build Koucai
• What works when using AI in an enterprise codebase with millions of LOC
• Is Claude being nice to me?

Connect