vegetarian-and-vegan-chef
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its documented research workflow. In
modules/research-checklist.md, the agent is instructed to conduct web searches and distill information from external sources to update its core guidance. 1. Ingestion points: External content fetched from web searches, Wikipedia, and Arxiv as instructed inmodules/research-checklist.md. 2. Boundary markers: Absent. The skill lacks instructions to use delimiters or warnings for the agent to ignore embedded instructions within retrieved data. 3. Capability inventory: The agent uses search and distillation tools to aggregate information and is tasked with modifying its core guidance based on that data. 4. Sanitization: Absent. No verification or sanitization steps are provided to validate external content before processing.
Audit Metadata