api-design-reviewer
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No instructions were found that attempt to bypass safety guidelines or override agent behavior.- [DATA_EXFILTRATION]: The skill does not perform network operations. It processes data locally and does not access sensitive system paths or credentials.- [REMOTE_CODE_EXECUTION]: No remote scripts are downloaded or executed. The provided Python scripts use only standard libraries.- [COMMAND_EXECUTION]: The skill includes local Python scripts for linting and scoring. These scripts process user-provided files deterministically and do not spawn shells or execute arbitrary commands.- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from OpenAPI specifications (JSON/YAML). While the analysis logic is deterministic, the output generated (such as summaries and documentation checks) may contain text from the input file which is then processed by the AI agent. This is a standard risk for tools processing external data.
- Ingestion points: Files like openapi.json are read by scripts/api_linter.py, scripts/breaking_change_detector.py, and scripts/api_scorecard.py.
- Boundary markers: Not present. The scripts interpret the JSON structure but do not wrap extracted text in safety delimiters.
- Capability inventory: The scripts can read from and write to the local filesystem (via output flags) and print to the console.
- Sanitization: Standard JSON parsing and regex-based convention checking are applied.
Audit Metadata