Skip to the collection

agent-native · Code Execution

Agent Sandbox

Listed on The Agent-Native Index

The trusted runtime for untrusted code

What it does

Agent Sandbox provides a managed, isolated runtime where AI agents can execute Python and shell commands, install dependencies, and exchange files through a single API. The product is positioned around giving each agent a personal remote computer and drive, with explicit support for sessioned execution and artifact pipelines.


Agent-facing interfaces

How agents onboard

https://agentsandbox.co/skill.md

See the full onboarding section in the dossier.

Protocol surface

Interface Detail
REST API https://agentsandbox.co/docs
Python SDK pip install agentsandbox-sdk
URL Onboarding https://agentsandbox.co/skill.md