Skills
Navigate
skills/aiagenta2z/onekey-agent-router/gemini-nano-banana/Gen Agent Trust Hub

gemini-nano-banana

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFECREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The files scripts/generate_image_gemini.py and scripts/generate_image_nano_banana.py contain a hardcoded demo key BETA_TEST_KEY_MARCH_2026 as a fallback for the DEEPNLP_ONEKEY_ROUTER_ACCESS environment variable.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting external data used in remote tool invocations and file path handling.
  • Ingestion points: The prompt, image_name, and output_folder parameters provided in the tool payload.
  • Boundary markers: No boundary markers or protective instructions are used to separate user data from agent instructions.
  • Capability inventory: The skill uses OneKeyAgentRouter.invoke to generate images and handles file system paths for saving the resulting images.
  • Sanitization: No input validation or path sanitization is performed on the output_folder or image_name parameters before being sent to the router service.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 08:45 AM
Security Audit — agent-trust-hub — gemini-nano-banana