Notebook

Warn

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides the ability to execute arbitrary code snippets through a local Node.js script located at /job/.pi/skills/notebook/notebook.js.
  • [COMMAND_EXECUTION]: Evidence: usage examples demonstrate running code directly from command-line arguments (e.g., node .../notebook.js cell "console.log('Hello')") and executing entire notebooks from local files (e.g., node .../notebook.js run notebook.json).
  • [PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection by processing external notebook.json files.
  • [PROMPT_INJECTION]: Ingestion points: notebook.json is read and processed as a notebook. Boundary markers: no delimiters or 'ignore' instructions are present in the skill definition. Capability inventory: the skill can execute arbitrary Node.js code and write/export results to the filesystem. Sanitization: no evidence of input validation or sanitization is documented for the processed notebook content.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 1, 2026, 05:11 AM