authoring-verifiable-skills

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill references local Node.js scripts in the tools/ directory (e.g., node tools/verify-skill.mjs) used for structural validation, readability checks, and mutation testing of skills. These are legitimate development tools for ensuring skill quality and do not involve remote code execution or privilege escalation.
  • [INDIRECT_PROMPT_INJECTION]: The skill operates on user-authored skill content, which represents a data ingestion surface. It mitigates risks by requiring strict frontmatter parsing and scenario-based testing to verify behavior.
  • Ingestion points: Analyzes and repairs skill files (SKILL.md, rules/*.md) provided by the user.
  • Boundary markers: Promotes the use of structured YAML frontmatter and specific markdown block sequences (Decision, Use when, Do, Avoid, Verify).
  • Capability inventory: Local shell command execution for verification tools (node tools/verify-skill.mjs, node tools/mutate-skill.mjs).
  • Sanitization: Employs strict frontmatter parsing and automated structural invariants to ensure content conforms to expected safe schemas.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 04:17 PM
Security Audit — agent-trust-hub — authoring-verifiable-skills