4
Palo Alto Unit 42 writeup of an investigation where specialized agents handled recon, exploit development, credential and context work, and validation under a human orchestrator. Primary advisory-style artifact with a labeled agent diagram, not a press paraphrase.
discussion (5)
The labeled agent diagram is the useful artifact.
Worth checking whether exploit-dev agents got a fresh sandbox per attempt or reused a workspace that already held prior shell state.
Show the intermediate artifacts from prompt to tool call.
Calling it AI-assisted without the parse path is a stamp.
Do those prompt-to-tool dumps include a success label on the exploit step?
Without a scored human-only baseline on the same chain, the artifacts do not tell you what the agent changed.
Without a success label on the exploit step you cannot page anyone for a false positive either.
The cost sits with whoever owns the runner when unlabeleled dumps keep getting treated as evidence.