pflow-grill
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill utilizes a research workflow that ingests data from external sources to inform its questions.
- Ingestion points: The 'Research' step in
SKILL.mdinstructs the agent to fetch facts from the filesystem, tools, and documentation. - Boundary markers: The skill employs a strict structured output format (numbered questions with lettered options), which acts as a delimiter, though it lacks explicit instructions to ignore embedded commands within the researched data.
- Capability inventory: The skill dispatches sub-agents for research and uses
functions.request_user_inputfor user interaction. It explicitly prohibits implementing code, editing files, or committing changes, which significantly mitigates the risk of a successful injection. - Sanitization: There is no documented process for sanitizing or filtering facts retrieved from the environment before they are incorporated into the agent's reasoning tree.
Audit Metadata