Oracle InfoGenius

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill utilizes a third-party Model Context Protocol (MCP) tool mcp__nanobanana__generate_image. The author or namespace nanobanana is not a recognized trusted source in the analyzer's framework, representing an unverifiable external dependency risk.
  • [PROMPT_INJECTION] (LOW): The skill is susceptible to Indirect Prompt Injection. Ingestion points: Web search results via the WebSearch function. Boundary markers: Absent; the search data is interpolated directly into the image generation prompt without delimiters. Capability inventory: Image generation via mcp__nanobanana__generate_image. Sanitization: Absent; there is no validation or escaping of external content before processing.
  • [DATA_EXFILTRATION] (LOW): Information disclosure of the local system environment. The skill hardcodes a specific Windows file path /mnt/c/Users/Frank/oracle-work/projects/deliverables/images/, which reveals a local username ("Frank") and an internal directory structure.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:46 PM