The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes uv run and bash to execute local scripts such as catalog_verify.py, catalog_validate.py, and drift.sh for API verification and catalog validation tasks.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data including vendor API documentation and OpenAPI specifications, which presents an attack surface for indirect prompt injection. 1. Ingestion points: Vendor facts intake in Step 0 and OpenAPI spec/documentation processing in Step 6. 2. Boundary markers: The skill lacks formal boundary delimiters for ingested external data, relying on agent procedural instructions and subsequent validation steps. 3. Capability inventory: The skill performs network operations (API probing) and executes local verification scripts. 4. Sanitization: Step 7 explicitly mandates scrubbing PII and sensitive credentials from all captured examples before committing them to the repository.

vendor-listing