grill-with-style
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill accesses local project files and the codebase to inform its interview process, which is a standard functionality but technically constitutes a data ingestion surface. \n- Ingestion points: The skill reads files from the
style-guide/directory and explores the codebase as instructed inSKILL.md. \n- Boundary markers: No specific delimiters or safety instructions are defined in the prompt to separate ingested file content from the agent's core instructions. \n- Capability inventory: The agent uses standard file reading and codebase exploration tools to gather context. \n- Sanitization: There is no evidence of specific sanitization or filtering applied to the content of the style guide files or code before it is processed by the agent.
Audit Metadata