Skills
Navigate
skills/bankai-skills/superpowers/nano-banana-2/Gen Agent Trust Hub

nano-banana-2

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references external installation instructions and related skills hosted on GitHub at raw.githubusercontent.com/inference-sh/.
  • [REMOTE_CODE_EXECUTION]: Instructions direct the user to install external components using npx skills add, which fetches and integrates additional functionality from the inference-sh repository.
  • [COMMAND_EXECUTION]: The skill requires the Bash tool to execute the belt CLI for logging in and running inference applications (e.g., belt app run).
  • [INDIRECT_PROMPT_INJECTION]: The skill features an attack surface for indirect prompt injection by ingesting untrusted data through the prompt parameter and images URL list.
  • Ingestion points: The prompt string and images array in the belt app run command (SKILL.md).
  • Boundary markers: None present; the user prompt is interpolated directly into the CLI input JSON.
  • Capability inventory: The skill can execute shell commands via the belt CLI (Bash tool).
  • Sanitization: None detected; user input is passed to the underlying model and CLI tool as structured JSON.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:54 PM
Security Audit — agent-trust-hub — nano-banana-2