comfy-model-onboarding
Pass
Audited by Gen Agent Trust Hub on May 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to use local CLI tools like
comfy-modelsandcomfy-imagegen, alongside standard filesystem commands likelsandfind, to manage model profiles and verify their functionality.\n- [EXTERNAL_DOWNLOADS]: The skill referencescomfy-model-downloaderto retrieve missing model files required for the supported profiles.\n- [PROMPT_INJECTION]: The skill processes data from the local filesystem, such as directory listings and file metadata, which creates a surface for indirect prompt injection.\n - Ingestion points: The agent reads file names and
safetensorsmetadata from the model directory (SKILL.md).\n - Boundary markers: No specific delimiters are used to separate external file data from the agent's instructions.\n
- Capability inventory: The agent can modify local configurations and execute various system tools.\n
- Sanitization: There is no mention of sanitizing or validating the content read from file metadata.
Audit Metadata