legal-canned-responses
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill was analyzed for 11 threat categories, including prompt injection, data exfiltration, and remote code execution. No malicious patterns or vulnerabilities were identified. All scripts were found to be localized text-processing utilities without network or filesystem access.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests raw inquiry text from users for classification and template filling. While this presents an untrusted data surface, the provided Python scripts lack dangerous capabilities such as network access, filesystem writes, or command execution. Boundary markers are absent in the scripts, but the low-capability environment minimizes the risk of successful exploitation.
Audit Metadata