learn
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides high-level reasoning instructions and does not contain any executable scripts, binary code, or hardcoded credentials.
- [PROMPT_INJECTION]: No direct prompt injection or behavior override commands were identified. The skill interacts with external content via WebFetch and WebSearch to build its mental models, creating an indirect ingestion surface. Ingestion points: SKILL.md (Step 1: Survey, Step 3: Explore). Boundary markers: None specified for external content. Capability inventory: Read, Grep, Glob, WebFetch, WebSearch. Sanitization: The instructions do not specify validation or escaping of retrieved content.
- [DATA_EXFILTRATION]: While the skill uses network tools, they are employed for the primary purpose of fetching information rather than sending sensitive local data to external endpoints.
Audit Metadata