setup-python-tools
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill facilitates the runtime execution of Python code in the browser via Pyodide. The instructions specify that the execution environment automatically fetches and runs code from a remote agent configuration when tools are invoked, which constitutes dynamic execution of externally sourced content.
- [INDIRECT_PROMPT_INJECTION]: The architecture described creates a vulnerability to indirect prompt injection by processing untrusted data (Python code from a remote config) that is executed in the user's browser context.
- Ingestion points: Code is fetched from the 'agent CDF config' and executed via the
usePyodideRuntimehook. - Boundary markers: The instructions do not provide or mandate specific delimiters or verification steps to validate the integrity of the remote code before execution.
- Capability inventory: The skill enables full client-side Python execution capabilities, including package installation (
requirements: ['pandas', 'numpy']) and interaction with the browser environment. - Sanitization: There are no sanitization or escaping protocols mentioned for the code fetched from the remote configuration prior to its execution in the Pyodide runtime.
Audit Metadata