diagnose-generation-failure
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- Indirect Prompt Injection (LOW): The skill processes untrusted external OpenAPI specifications, creating a surface for indirect prompt injection. 1. Ingestion points: File paths for OpenAPI specs provided as input and error output strings. 2. Boundary markers: Absent; there are no instructions to the agent to ignore instructions embedded within the spec files. 3. Capability inventory: Shell execution of 'speakeasy', 'yq', and 'jq' for linting and parsing. 4. Sanitization: Absent; the skill does not specify validation or sanitization of the spec content before processing. The risk is categorized as low as the operations are diagnostic.- Command Execution (SAFE): The skill instructs the agent to execute 'speakeasy', 'yq', and 'jq' commands. These are legitimate utilities for the stated purpose of diagnosing SDK generation issues and do not exhibit malicious patterns or privilege escalation.
Audit Metadata