professional-communication
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (LOW): An attack surface exists as the skill utilizes Read, Glob, and Grep tools to process external file content. 1. Ingestion points: File contents accessed via Read and Grep. 2. Boundary markers: Absent in the skill instructions. 3. Capability inventory: No file-write, command execution, or network operations are permitted; capabilities are limited to reading. 4. Sanitization: None identified. The risk is limited to influencing the agent's internal reasoning or output text.
- [SAFE] (SAFE): No evidence of hardcoded credentials, obfuscation, persistence mechanisms, or malicious command execution patterns was found in the skill content.
Audit Metadata