skills/cosmix/loom/loom-testing/Gen Agent Trust Hub

loom-testing

Pass

Audited by Gen Agent Trust Hub on May 8, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: No override instructions, safety bypasses, or system prompt extraction attempts were found. The instructional tone is standard for a technical guide.
  • [DATA_EXFILTRATION]: No access to sensitive environment variables or system paths was detected. Code examples use clearly marked placeholders like 'test_key' for API keys, which is standard testing practice.
  • [REMOTE_CODE_EXECUTION]: The skill references only legitimate, well-known testing frameworks (e.g., pytest, jest, vitest, mocha). There is no evidence of downloading or executing scripts from untrusted remote sources.
  • [COMMAND_EXECUTION]: Bash usage is scoped to standard development and testing workflows such as executing test runners or infrastructure validation tools.
  • [DATA_EXPOSURE]: While the skill describes processing external data (e.g., CSV fixtures, Terraform plans), it emphasizes validation and correctness without introducing insecure handling patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
May 8, 2026, 05:52 AM
Security Audit — agent-trust-hub — loom-testing