testing-llm

Pass

Audited by Gen Agent Trust Hub on Mar 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides comprehensive patterns for testing AI and browser integrations using industry-standard frameworks such as DeepEval and Playwright.
  • [PROMPT_INJECTION]: A string containing 'Ignore previous instructions' was identified in examples/llm-test-patterns.md. Detailed analysis confirms this is a benign code example for a negative test case (verifying the system under test resists injection attempts) rather than a malicious instruction targeting the agent.
  • [CREDENTIALS_UNSAFE]: The skill documents and promotes secure secret management, specifically instructing users to avoid live API calls in CI, use environment variables, and implement header filtering for sensitive data (e.g., Authorization, x-api-key) when using VCR.py recordings.
  • [EXTERNAL_DOWNLOADS]: The skill references installation of well-known and trusted testing frameworks including DeepEval, RAGAS, and Playwright via standard package managers (pip, npm).
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 26, 2026, 10:46 AM