nano-banana-edit
Pass
Audited by Gen Agent Trust Hub on Aug 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
runcomfycommand-line tool to perform image editing tasks. It passes instructions and parameters using a structured JSON string via the--inputflag, which mitigates traditional shell injection vulnerabilities. - [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
@runcomfy/clipackage from the NPM registry and fetches skill definitions from a GitHub repository (agentspace-so/runcomfy-skills). These are standard operational requirements for integrating with the RunComfy platform. - [INDIRECT_PROMPT_INJECTION]: The skill processes external image URLs provided by the user. The documentation identifies image-based prompt injection as an inherent risk of model processing and notes that the CLI environment provides basic boundaries (JSON-only transport) to reduce local execution risks.
- [DATA_EXPOSURE]: The documentation explains how the RunComfy CLI manages API tokens (storing them in
~/.config/runcomfy/token.jsonwith 0600 permissions) and recommends using theRUNCOMFY_TOKENenvironment variable for secure automation in CI/CD environments.
Audit Metadata