review

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests and processes untrusted data from inbound legal agreements (provided via files, links, or text) to make routing decisions and generate review memos.
  • Ingestion points: Document titles and body content from file paths, Google Drive links, CLM IDs, or pasted text.
  • Boundary markers: There are no defined delimiters or instructions to treat the ingested document content as untrusted data, which could allow a maliciously crafted document to influence the agent's logic.
  • Capability inventory: The skill reads local configuration files from ~/.claude/plugins/config/claude-for-legal/commercial-legal/CLAUDE.md and invokes other internal skills (e.g., nda-review, vendor-agreement-review).
  • Sanitization: No explicit sanitization or validation of the input text is performed before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 03:57 PM
Security Audit — agent-trust-hub — review