Post #2072438

What's new? I built this pentesting and red/purple teaming llm skills-based assistant. It's pretty neat and learned a lot (was a learning/research project for work that encompass about 100 separate skills for things I do using real code and data as concrete backing it uses). Using agent skills and hooks are way more flexible and powerful than I ever thought. It's sorta difficult to go back to raw now. With a good infra and a ton more tokens, I see how determined can operationalize this.