ramp

Pass

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: SAFE
Full Analysis
  • Local File Access: The skill reads from ~/.claude/plugins/config/claude-for-legal/legal-clinic/CLAUDE.md to personalize the onboarding experience. This is a configuration file containing clinic metadata, such as practice areas and jurisdiction, and is accessed within the plugin's expected configuration scope.
  • Promoting Best Practices: The instructions include clear directives for the AI to emphasize that its outputs are starting points rather than final work products, encouraging critical review and citation verification by the student.
  • Controlled Training Environment: The skill uses hypothetical scenarios for practice exercises, ensuring that initial tool usage occurs in a low-stakes environment. No evidence of unauthorized data exfiltration, remote code execution, or privilege escalation was found.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 12, 2026, 09:46 PM
Security Audit — agent-trust-hub — ramp