ard-registry-builder
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a legitimate developer tool for managing ARD metadata.
- [EXTERNAL_DOWNLOADS]: The validation and testing tools fetch remote manifests and probe API endpoints. These network operations are strictly limited to the skill's documented purpose and target user-specified URLs.
- [REMOTE_CODE_EXECUTION]: The skill utilizes the well-known jsonschema Python package for structural validation. All logic is implemented in clear, standard Python code without dynamic execution of untrusted strings.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external manifests and registry responses. While these are ingestion points for untrusted data, the risk is mitigated by structural JSON parsing and the skill's specific role as a validator. Ingestion points: Remote manifest URLs and local files processed by scripts/validate_catalog.py and scripts/test_registry.py. Boundary markers: None explicitly used in script output to the agent. Capability inventory: File system writes via scripts/new_catalog.py and network GET/POST requests via testing scripts. Sanitization: All input is processed as structured JSON; semantic validation checks for URN formats and schema compliance.
Audit Metadata