test-health

Pass

Audited by Gen Agent Trust Hub on Apr 20, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill can execute project-defined test and coverage commands (e.g., npm run coverage, pytest) when the --collect flag is explicitly used in Full Mode. This is a standard and documented feature of the development utility.
  • [DATA_EXPOSURE]: Historical test metrics are persisted in the .claude/cache/test-health/ directory. This is used solely for computing trend deltas and remains within the local project environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data including test runner output (stdout) and coverage artifacts (LCOV, XML, JSON).
  • Ingestion points: Reads coverage files and test runner output from the repository environment.
  • Boundary markers: Data is parsed into structured metrics and presented in Markdown tables; the parsing logic itself acts as a filter.
  • Capability inventory: Includes execution of restricted bash commands (git, npm, cargo, etc.) and the ability to trigger other agent skills like /check-coverage.
  • Sanitization: Extraction logic uses specific regular expressions and strict JSON/numeric parsing to convert raw text into quantitative metrics, effectively neutralizing potential embedded instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 20, 2026, 07:24 PM
Security Audit — agent-trust-hub — test-health