E2B gives AI agents isolated cloud sandboxes in which to run AI-generated code, process data, and use tools. Each sandbox is a Linux VM created on demand, and you start and control sandboxes from the JavaScript or Python SDK.
A template defines the environment a sandbox starts with. Pausing a sandbox saves both its filesystem and its memory, and paused sandboxes are kept until you kill them, so an agent can resume where it stopped.
Two add-on SDKs build on the base sandbox: Code Interpreter runs code snippets and returns their results, and Desktop exposes mouse, keyboard, screenshot, application, and desktop streaming APIs for computer-use agents. E2B runs as a hosted service, and the runtime can be self-hosted with Terraform on AWS, Google Cloud, Azure, or a general Linux machine.
Features
- Sandboxes: Linux VMs created on demand for an agent, controlled from JavaScript or Python
- Templates: define the environment each sandbox starts with
- Pause and resume: pausing saves filesystem and memory, and paused sandboxes persist until killed
- Code Interpreter: an SDK that runs code in a sandbox and returns execution results
- Desktop: mouse, keyboard, screenshot, application, and desktop streaming APIs for computer-use agents
- GitHub Actions: sandboxes in CI workflows for testing, validation, and AI code reviews
- Agent-readable docs: a public docs MCP server and an llms.txt index
- Self-hosting: Terraform deployment on AWS, Google Cloud, Azure, or a general Linux machine
