gpt-image-2
Pass
Audited by Gen Agent Trust Hub on Apr 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
runcomfyCLI tool to perform image generation and editing tasks, passing user-controlled prompt data to the tool as a structured JSON string. - [EXTERNAL_DOWNLOADS]: The documentation instructs the user to install the
@runcomfy/clipackage from the public npm registry and usesnpxto fetch skill configurations from the vendor's repository. - [PROMPT_INJECTION]: The skill provides an interface for processing user-supplied prompts and image URLs through a remote model, creating a surface for indirect prompt injection.
- Ingestion points: User-provided strings and image URLs passed into the
--inputargument of theruncomfycommand inSKILL.md. - Boundary markers: Commands utilize JSON to delimit user input from the rest of the execution parameters.
- Capability inventory: The skill can initiate network requests to the RunComfy Model API and write output files to a specified directory on the local file system.
- Sanitization: The documentation notes that the CLI tool transmits data over HTTPS and does not perform shell expansion on the prompt contents.
Audit Metadata