agent-testing-harness

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a guide for implementing testing pyramids, unit tests, and integration tests for AI agents using the Vitest framework. No executable malicious code or dangerous commands are present within the skill content.
  • [SAFE]: Secret management within the provided code examples adheres to security best practices by utilizing environment variables (process.env.TEST_API_KEY) and CI/CD secrets rather than hardcoded credentials.
  • [SAFE]: The skill includes defensive programming examples, such as a test case designed to verify if an agent can identify a code injection vulnerability (eval(userInput)), which promotes secure development practices.
  • [SAFE]: All network-related code snippets target well-known services (e.g., OpenAI API) for legitimate testing purposes and do not exhibit exfiltration patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 01:19 PM
Security Audit — agent-trust-hub — agent-testing-harness