Reviewer Pattern — Skills

Pattern: Reviewer
Component: skills.md
Version: 1.3 | Updated: 2026-07-29

---

Skill	Path	Why required
Code review	skills/code-review.skill.md	Severity-calibrated findings humans can own
Security	skills/security.skill.md	AuthZ, injection, secrets, tool allowlists
Debug	skills/debug.skill.md	When findings need repro before Block

Related agents

• Reviewer agent
• Security agent

Skill activation triggers

Run Code review on every AI-touched PR. Escalate to Security when AuthZ, secrets, injection, or tool allowlists change. Escalate to Debug only when a finding needs a failing repro before severity can be set.

Verification that skills worked

• Each finding cites path:line a human can open
• Critical/High security findings block merge until owned
• Approve is never issued solely because CI is green

Changelog

• 1.3 (2026-07-29): Reviewer skill activation and verification rewritten for this envelope.