refine
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external workflow data, which represents a potential surface for instructions embedded in data to influence the agent. 1. Ingestion points: Workflow context retrieved via the /agent-workflow preparation step in SKILL.md. 2. Boundary markers: The instructions lack specific delimiters or warnings to ignore malicious content within the analyzed workflow. 3. Capability inventory: The skill focus is on analysis and text-based feedback; it does not directly invoke high-risk system or network tools. 4. Sanitization: No explicit validation or filtering of the workflow content is mentioned before processing.
- [SAFE]: The skill advocates for proactive security measures, including the mandatory redaction of PII (Personally Identifiable Information) from logs and the use of configuration files rather than hardcoded credentials.
Audit Metadata