flux-2-klein
Pass
Audited by Gen Agent Trust Hub on May 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Directs the user to install the
@runcomfy/clipackage from the npm registry as a prerequisite for using the skill. - [COMMAND_EXECUTION]: Executes the
runcomfycommand-line utility to interact with the RunComfy Model API, involving subprocess calls and network operations to authorized RunComfy domains. - [PROMPT_INJECTION]: The skill processes untrusted user data which could contain malicious instructions for the underlying AI model.
- Ingestion points: User-provided text prompts and external reference image URLs passed via CLI arguments in SKILL.md.
- Boundary markers: The CLI uses JSON-formatted input structures to transmit data, providing structural separation between the command and the payload.
- Capability inventory: Subprocess execution of the
runcomfybinary and file system writes to a user-specified output directory. - Sanitization: Relies on the RunComfy CLI's internal handling and HTTPS transmission; the skill explicitly documents the risks of image-based prompt injection in its security section.
Audit Metadata