takedown

Pass

Audited by Gen Agent Trust Hub on May 17, 2026

Risk Level: SAFE
Full Analysis
  • Local File Access: The skill is designed to read a configuration file located at ~/.claude/plugins/config/claude-for-legal/ip-legal/CLAUDE.md. This is used to load the user's legal practice profile, enforcement posture, and approval matrix. Accessing these files is necessary for the skill to provide contextually relevant legal drafting assistance.
  • External Data Processing (Indirect Prompt Injection Surface): In --respond mode, the skill processes incoming takedown notices provided by the user (e.g., PDF or text files). While processing untrusted external text can be a surface for indirect instructions, the skill mitigates this by using a highly structured triage process and enforcing 'Loud Gates' that require explicit human review and attorney oversight before any output is finalized.
  • Controlled Output Environment: The skill writes its drafts and triage memos to a specific local directory structure (~/.claude/plugins/config/claude-for-legal/ip-legal/matters/). This ensures that legal work product is organized and stored in a predictable location according to the user's matter-management settings.
  • Absence of Network Exfiltration: The analysis confirmed that the skill does not perform automated network requests or data transmission. All drafting is done locally, and the instructions explicitly state that the user is responsible for submitting any final notices through the service provider's official channels.
  • Citation and Case Law Verification: The skill includes instructions to source-tag and verify legal citations using external research tools, which helps prevent the inclusion of inaccurate or halluncinated legal information in drafts.
Audit Metadata
Risk Level
SAFE
Analyzed
May 17, 2026, 03:59 AM
Security Audit — agent-trust-hub — takedown