comfyui-gateway
Pass
Audited by Gen Agent Trust Hub on Apr 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill documentation describes a legitimate and well-architected utility for managing ComfyUI workflows and jobs.
- [CREDENTIALS_UNSAFE]: The provided documentation and integration examples use clearly labeled placeholders (e.g., 'your-api-key', 'sk-admin-key') and default credentials for services like MinIO ('minioadmin'). No actual sensitive data or real-world credentials are exposed.
- [EXTERNAL_DOWNLOADS]: References to external resources are limited to standard package managers (npm) and well-known container registries (ghcr.io). These are standard practices for the stated purpose of the skill.
- [COMMAND_EXECUTION]: Shell commands included in the documentation are restricted to standard development tasks such as project initialization, server execution, and network troubleshooting (e.g., ufw, iptables). No dangerous or unauthorized command execution patterns were found.
- [PROMPT_INJECTION]: The skill includes an assessment of Indirect Prompt Injection risks. Ingestion points: User-supplied 'inputs' in the /jobs endpoint (SKILL.md). Boundary markers: '{{placeholder}}' template syntax (SKILL.md). Capability inventory: Network requests to ComfyUI and file storage (SKILL.md). Sanitization: Zod schema validation and 'Privacy Mode' log redaction (SKILL.md).
- [DATA_EXFILTRATION]: While the gateway transmits data to ComfyUI and external storage (S3), these operations are core to its documented functionality and are protected by authentication and domain allowlists. There are no patterns indicating unauthorized data harvesting.
Audit Metadata