comfyui-mcp-agent

Warn

Audited by Gen Agent Trust Hub on Jul 18, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches configuration and resources from external sources including HuggingFace, Civitai, and GitHub repositories. Examples include model checkpoints and custom node repositories like github.com/ltdrdata/ComfyUI-Manager.git.
  • [REMOTE_CODE_EXECUTION]: Includes features to install custom nodes via tools like install_custom_nodes, which automate cloning repositories and installing dependencies through pip or npm. This allows for the installation of code from arbitrary remote sources.
  • [COMMAND_EXECUTION]: Executes system commands such as npx, python, and git for setup, service management, and workflow automation.
  • [PROMPT_INJECTION]: Processes external data in workflow JSONs and prompts (e.g., in enqueue_workflow). No boundary markers or sanitization filters are specified for these inputs while the skill has access to administrative tools like software installation and process management.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 18, 2026, 12:32 AM
Security Audit — agent-trust-hub — comfyui-mcp-agent