Tag
1 entry
Designing an LLM agent for security work that can look at a lot and do almost nothing on its own, with typed tools, deterministic scoring, human approval and an audit trail.