impact-measurement

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill operates as a purely analytical tool, focusing on auditing existing codebases for program evaluation rigor (logic models, indicator tracking, etc.).
  • [SAFE]: It uses standard file discovery patterns (searching for package.json, requirements.txt, etc.) to understand the technology stack of the software being analyzed.
  • [SAFE]: The instructions emphasize ethical data collection and beneficiary privacy, specifically instructing the agent to check for informed consent and PII minimization.
  • [SAFE]: While it includes a telemetry section, it is restricted to logging execution metadata (success/failure) to a local directory (~/.claude/projects/) only if it already exists, which is a common pattern for skill improvement and does not involve external data transmission.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 10:56 AM
Security Audit — agent-trust-hub — impact-measurement