comfyui
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates on external data formats such as ComfyUI JSON workflows and user-controlled custom task directories, presenting a functional indirect prompt injection surface.
- Ingestion points: Evaluates JSON workflow files stored in
comfyui_workflows/and custom task-specific subdirectories. - Boundary markers: Utilizes 'bounded live-schema authoring' and strictly filters the LLM-facing control surface to approved MCP verbs, effectively isolating the agent from arbitrary execution.
- Capability inventory: Includes generation execution, lifecycle management, and asset handling via the
story_comfytool, rather than through direct shell commands. - Sanitization: Employs a managed official control runtime and deterministic multi-job validation for generations, ensuring that batch executions remain within defined bounds.
Audit Metadata