Security Policy • AI Agent SDK

SecurityPolicy: Security policy for sandbox execution.

SecurityPolicy

Defined in the config module.

Security policy for sandbox execution.

Attributes: allow_network: Whether to allow network access allow_file_write: Whether to allow file writes allow_subprocess: Whether to allow subprocess creation allowed_paths: List of paths that can be accessed blocked_paths: List of paths that are blocked allowed_commands: List of allowed shell commands (empty = all) blocked_commands: List of blocked shell commands allowed_imports: List of allowed Python imports (empty = all) blocked_imports: List of blocked Python imports max_output_size: Maximum output size in bytes

Properties

Methods

Source