grammarly-api-reliability

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is designed for metadata-only analysis. The analyze_job_receipt.py script recursively scans input JSON and rejects any entry containing forbidden substrings like 'token', 'header', 'body', or 'content', preventing accidental processing of credentials or private document data.
  • [COMMAND_EXECUTION]: The script analyze_job_receipt.py is executed via python3 but contains no logic for spawning subprocesses or executing shell commands. It relies entirely on standard library modules for JSON parsing and argument handling.
  • [PROMPT_INJECTION]: The evaluation specification eval-spec.yaml contains example prompt injection strings. These are clearly delineated as test cases for adversarial validation and do not influence the skill's runtime instructions.
  • [EXTERNAL_DOWNLOADS]: The skill includes links to official Grammarly API documentation for reference. No automated downloads or execution of remote scripts are performed.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:43 AM
Security Audit — agent-trust-hub — grammarly-api-reliability