nano-banana-2
Pass
Audited by Gen Agent Trust Hub on Aug 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides command templates for interacting with the
runcomfyCLI tool to perform model runs locally. - [EXTERNAL_DOWNLOADS]: The skill specifies downloading generated content from official RunComfy domains (
*.runcomfy.netand*.runcomfy.com). - [PROMPT_INJECTION]: The skill includes an attack surface for indirect prompt injection via user-supplied content.
- Ingestion points: Untrusted text prompts and media resource URLs are ingested through the CLI's
--inputparameter as shown in the fileSKILL.md. - Boundary markers: The instructions recommend passing data via JSON-formatted strings, which helps prevent shell command injection by encapsulating the user input.
- Capability inventory: The skill utilizes the
runcomfyCLI which has the capability to perform network requests and write files to the local file system. - Sanitization: There is no evidence of input validation or sanitization within the skill instructions, though it notes that the CLI itself avoids shell expansion.
Audit Metadata