nanobanana-visual

Pass

Audited by Gen Agent Trust Hub on Apr 26, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill transmits content derived from the local outputs/brief.md file to an external API endpoint (https://api.nanobanana.com/v3/generate). This is the core functionality of the skill but involves sending user-supplied project data to a third-party service.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data from outputs/brief.md without sanitization or boundary markers.\n
  • Ingestion points: The skill reads outputs/brief.md in the first step of its workflow to extract core messages and visual guides.\n
  • Boundary markers: There are no delimiters or instructions to ignore embedded commands within the processed brief file.\n
  • Capability inventory: The agent has the capability to perform network POST requests via the requests library and write image files to the local file system (outputs/visuals/).\n
  • Sanitization: No sanitization, escaping, or validation is performed on the text extracted from the brief before it is used to construct prompts for the image generation API.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 26, 2026, 08:51 AM