NewsShow HN: Scan your AI agents for dangerous capabilitiesJuly 6, 20261 min readSource: Hacker NewsHN score 32,comments 17