audit-skill-conflicts

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is designed for internal consistency auditing of an agent's skill pack. It primarily involves reading markdown files (SKILL.md, commands/*.md) and rules (rules/*.mdc) to identify semantic overlaps or contradictory instructions.
  • [SAFE]: No network operations, remote script downloads, or credential-access patterns were detected. The skill operates locally on the project's documentation and configuration files.
  • [SAFE]: The 'Phase 0 — Load the corpus' step uses standard file enumeration patterns to index documentation. While it searches for 'highest-risk' .mdc rules, this is to ensure they don't override other skills, which is a defensive/safety-oriented behavior.
  • [SAFE]: The skill explicitly instructs the agent to 'Rewrite nothing until approved,' maintaining a low-privilege, user-controlled workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 05:55 AM
Security Audit — agent-trust-hub — audit-skill-conflicts