nb-image-generation

Warn

Audited by Gen Agent Trust Hub on Mar 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill facilitates the execution of the nanobanana (alias nb) CLI tool to process image generation and manipulation requests.
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the @giorgioliapakis/nanobanana global NPM package. This package is owned by the skill's author and is the primary dependency for the skill's functionality.
  • [DATA_EXFILTRATION]: The skill's documentation in references/editing.md specifies that the CLI tool automatically searches for input files in sensitive local directories, including ~/Downloads/ and ~/Desktop/. This automated access to common user storage locations increases the risk of unintended file exposure.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted image data from various local and external sources.
  • Ingestion points: Files are ingested from the current directory, ./images/, ./input/, ./nanobanana-output/, ~/Downloads/, and ~/Desktop/.
  • Boundary markers: The system lacks explicit boundary markers or instructions to the model to ignore potential malicious prompts embedded within image data.
  • Capability inventory: The skill has the capability to execute shell commands and write output files to the local file system.
  • Sanitization: There is no evidence of content validation or sanitization of the image files before they are processed by the underlying AI models.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 17, 2026, 10:57 AM