draft

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFE
Full Analysis
  • [Local Configuration Access]: The skill reads jurisdiction settings and document templates from the ~/.claude/plugins/config/claude-for-legal/ directory. This is standard behavior for retrieving the localized rules and practice area guidelines necessary for the skill's primary function.
  • [Processing External Data]: The skill populates documents using information from 'case notes' or 'intake summaries' (Ingestion points). This ingestion process represents a surface where external content could potentially influence the agent. The skill manages this by limiting its functionality to text generation (Capability inventory) and implementing a robust verification process. This includes inline flags like [VERIFY] for assertions and a mandatory supervision workflow (Sanitization/Boundary markers) involving student checklists and supervising attorney sign-off before any draft is finalized.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 01:04 AM
Security Audit — agent-trust-hub — draft