policy-monitor

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill performs legitimate automated compliance monitoring. No malicious patterns or security risks were identified.- [COMMAND_EXECUTION]: The skill suggests periodic execution via external scheduling tools (Calendar, CI), but does not execute system commands to establish persistence itself.- [DATA_EXPOSURE]: Accesses plugin-specific configuration files in ~/.claude/plugins/config/claude-for-legal/ and legal documents from a user-defined directory. These operations are restricted to the intended functional scope.- [PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection as it ingests untrusted legal documentation without using explicit boundary markers.
  • Ingestion points: Reads Privacy Impact Assessments (PIAs), Data Processing Agreements (DPAs), and triage files from a configured output folder.
  • Boundary markers: Absent. The content is parsed and compared directly against existing policy commitments.
  • Capability inventory: File system read/write access (reading documents, updating scan dates in configuration).
  • Sanitization: No sanitization of ingested document content is performed before analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 07:06 AM
Security Audit — agent-trust-hub — policy-monitor