python-executor
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill facilitates the execution of arbitrary Python code on the inference.sh platform via the
beltCLI. - Ingestion points: Untrusted instructions could enter via the
codeparameter in the input schema withinSKILL.md. - Boundary markers: No explicit delimiters are used to separate user-provided data from code instructions within the examples.
- Capability inventory: The environment supports full Python 3.10 execution, network requests (via
requests,httpx), and file system writes to theoutputs/directory. - Sanitization: The skill relies on external server-side sandboxing provided by the inference.sh platform rather than local sanitization.
- [COMMAND_EXECUTION]: The instructions direct the agent to use the
beltCLI for logging in and running applications (belt login,belt app run). - [EXTERNAL_DOWNLOADS]: The documentation references external configuration and installation scripts hosted on GitHub (
raw.githubusercontent.com/inference-sh/skills/).
Audit Metadata