api-reference-quality
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, specifically machine-readable API specifications (OpenAPI, AsyncAPI, GraphQL SDL) and gateway route definitions. A maliciously crafted specification could contain instructions within descriptions or metadata intended to manipulate the agent's audit findings.
- Ingestion points: Untrusted specifications are ingested in Step 1 (Census the surface) and Step 2 (Audit every endpoint) as described in
SKILL.md. - Boundary markers: The skill does not provide explicit delimiters or instructions to the agent to ignore instructions embedded within the data being audited.
- Capability inventory: The skill provides logical instructions for comparison and scoring but does not include scripts for execution or automated network requests.
- Sanitization: There are no instructions for sanitizing or escaping the content of the ingested specifications.
- [DATA_EXFILTRATION]: The auditing workflow instructs the agent to examine sensitive data sources to identify undocumented endpoints, including gateway logs, SDK internals, and support tickets. While appropriate for a security/quality audit, this pattern involves processing potentially sensitive infrastructure information.
Audit Metadata