nano-banana-prompting
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied data, including reference images and existing prompts, to construct instructions for an image generation agent. 1. Ingestion points: User-provided reference images and previous prompts in Step 1, and answers to clarification questions in Step 2. 2. Boundary markers: The skill does not define explicit delimiters or instructions to ignore embedded commands when interpolating user data into the generated prompt in Step 4. 3. Capability inventory: The skill's primary capability is invoking the
nano-bananaimage generation skill; it does not perform file-writes, network operations, or code execution. 4. Sanitization: No sanitization or filtering of user-provided text is documented.
Audit Metadata