agent-teams-simplify-and-harden

SUSPICIOUS: The skill’s core behavior is broadly consistent with its stated purpose of orchestrated implementation and audit loops, so there is no clear evidence of malware or credential theft. The main risks are proportionality and trust: it installs a third-party personal-repo skill, enables broad autonomous code-edit/build/test loops with `bypassPermissions`, and chains into another skill, creating a medium-to-high operational security risk even though the workflow itself is plausible.