qa-agent-testing

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a guide for designing evaluation suites and does not contain any executable code that performs network operations, file system modifications, or unauthorized data access.
  • [SAFE]: It explicitly incorporates security-positive recommendations, such as testing for prompt injection, implementing tool sandboxing, and treating external tool outputs as untrusted.
  • [SAFE]: No evidence of obfuscation, hardcoded credentials, or data exfiltration was found in the documentation or metadata.
  • [SAFE]: The skill mentions a local Python script (scripts/score_suite.py) for scoring, which is a standard utility function for a testing harness.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 04:22 AM