Cloudflare's security-audit skill is at 11,882 stars for an adversarial six-phase audit loop.

cloudflare/security-audit-skill reached the HN front page September 17, though the repo dates to June 18. The skill runs reconnaissance to map architecture and trust boundaries, coverage-led hunting with isolated hunters, candidate validation where fresh verifiers try to disprove each finding, JSON output split into confirmed/needs_validation/rejected, independent record verification by new agents, and target-neutral reporting. The reusable idea for anyone writing skills is the adversarial validation loop plus the three-way verdict, which is what kept Cloudflare's fleet-wide audit from drowning in false positives across 128 repos.