vendor-ai-review
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's operations are restricted to its internal plugin configuration and matter-specific data paths within
~/.claude/plugins/config/claude-for-legal/. - [SAFE]: No network operations, external downloads, or remote code execution patterns were identified.
- [PROMPT_INJECTION]: The skill processes untrusted external data (vendor contracts), which constitutes an indirect prompt injection surface.
- Ingestion points: User-provided vendor terms (PDF attachments or pasted text).
- Boundary markers: None explicitly defined to isolate document content from instructions.
- Capability inventory: Read/write access to local plugin files; no network or subprocess capabilities available.
- Sanitization: Not described.
- Assessment: The risk is low because the skill's output is purely informational (analysis and redlines) and it lacks the capabilities to perform sensitive actions if an injection occurs.
Audit Metadata