fengtang-perspective
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and reference materials are focused on role-play and providing advice using specific mental models. No malicious commands, exfiltration patterns, or persistence mechanisms were found.
- [PROMPT_INJECTION]: The skill implements an agentic protocol that fetches untrusted data from the web via search tools (Ingestion points: SKILL.md Step 2). It attempts to isolate this data by creating an internal summary (Boundary markers: SKILL.md Step 2). The workflow utilizes the agent's web search capabilities (Capability inventory). No specific data sanitization rules are included in the skill body (Sanitization). This represents a low-risk operational surface consistent with the skill's primary function.
Audit Metadata