The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill is explicitly designed to execute arbitrary Python code on a remote server provided by the vendor. It utilizes the belt CLI to transmit and run code payloads.
[COMMAND_EXECUTION]: The skill requires access to the Bash tool to execute belt commands, which is used to interact with the remote Python execution environment.
[EXTERNAL_DOWNLOADS]: The skill references installation instructions for the belt CLI utility from the vendor's GitHub repository (github.com/inference-sh). This is part of the standard setup for the service.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection.
Ingestion points: The code input parameter accepts strings of Python code constructed by the agent.
Boundary markers: There are no boundary markers or instructions to sanitize or validate the content of the code block before execution.
Capability inventory: The remote environment supports network operations (via requests, selenium, etc.) and file outputs, allowing for complex multi-stage operations.
Sanitization: The skill relies entirely on the remote environment's sandboxing (as described in the documentation) rather than input-level sanitization.

python-executor