agency-evaluation-criteria

Pass

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or data exfiltration attempts were detected. The skill file is strictly instructional and defines assessment criteria for an AI agent.
  • [NO_CODE]: The skill contains no scripts, binaries, or executable command sequences. It defines requirements for external tools like Playwright and Lighthouse but does not implement them directly.
  • [PROMPT_INJECTION]: The skill's intended workflow involves evaluating untrusted data (external applications and project documents), creating a surface for indirect prompt injection.
  • Ingestion points: Built application URLs/paths and BRIEF documents.
  • Boundary markers: None present in the criteria documentation.
  • Capability inventory: Requirements for Bash and automated browser testing (Playwright).
  • Sanitization: No sanitization logic described in the assessment criteria.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 7, 2026, 06:35 PM