Security Auditor — Meta-Skill

The foundational security layer of the Skill ecosystem. It operates across three independent but interlocking modes. Decide which mode to run based on the context—or adhere to the user's explicit instructions.

Execution Decision Matrix:

• "I wrote a new skill, is it safe?" → Audit + Trust
• "This skill will be deployed to production" → Audit + Trust + Synthesize Guard rules
• "Scan the Ecosystem" → Audit + Trust on all skills
• Suspicious runtime behavior → Guard mode intervention

MODE 1 — AUDIT (Static Analysis)

Diagnose security flaws by deeply parsing a SKILL.md file. Assess risks exclusively through the skill's instructions without executing code.

Audit Steps