skill-catalog-builder
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill processes SKILL.md files from external discovery scopes, creating a surface for indirect prompt injection via skill descriptions. This is mitigated by the implementation.
- Ingestion points: scripts/build_catalog.py reads files from directory paths provided as arguments.
- Boundary markers: The generated catalog is returned as structured JSON, and descriptions are truncated to a maximum length.
- Capability inventory: The script is restricted to read-only filesystem access and lacks network or write capabilities.
- Sanitization: The skill implements manual frontmatter parsing to avoid YAML exploits, enforces an 8KB size limit, validates names with regex, and rejects symlinks.
Audit Metadata