comfy-model-onboarding
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process data from external sources which could be manipulated to influence agent behavior.
- Ingestion points: The skill performs metadata reads from
safetensorsfiles and inspects user-provided file paths (SKILL.md, Step 1 in Onboarding Flow) to identify architectures and select profiles. - Boundary markers: No explicit instructions are provided to the agent to treat metadata content as untrusted or to use specific delimiters when parsing file information.
- Capability inventory: The agent has the capability to execute shell commands via
uv runand perform file system operations likelsandfind(SKILL.md, Setup and Onboarding Flow sections). - Sanitization: There is no mention of sanitizing or validating the content of the
safetensorsmetadata before processing.
Audit Metadata