Self-hosted infrastructure that runs untrusted code in lightweight VM sandboxes, managed through a CLI, HTTP API and Python SDK.
- Stars
- 1,523
- Forks
- 122
- Open issues
- 97
- Last commit
- 11 hours ago
What is it?
SPACE (Sandboxed Platform for Agentic Code Execution) gives applications a separate Linux environment for running code. Each sandbox is an instance of a template with its own virtual machine and writable filesystem, and several commands can run inside the same sandbox. Applications call the HTTP API to create sandboxes, start processes, read their output and exit status, and delete them when finished.
The project targets cases where an AI agent, a user workspace, or a code-running job needs isolation from the host. The application decides what to run and which results to keep, while SPACE handles the sandbox lifecycle and the connection that carries command output. Lifecycle operations are asynchronous and return an operation ID to poll. Sandboxes can be kept, paused, suspended, and woken later.
SPACE is implemented in Rust and deploys to a local Kind Kubernetes cluster for development, with an optional single-node snapshot mode and warm pools for faster creates. It also includes an E2B compatibility wrapper for the E2B Python SDK.
Key features
- 01Creates sandboxes from templates, each with its own VM and writable filesystem.
- 02Runs multiple commands in one sandbox and streams their output.
- 03Manages lifecycle with keep, pause, suspend, wake, and delete operations.
- 04Exposes everything over HTTP and server-sent events, with an OpenAPI reference.
- 05Offers an E2B compatibility wrapper for the E2B Python SDK.
Use cases
- Running model-generated code for an AI agent in an isolated environment.
- Providing each user a separate code workspace on a shared platform.
- Executing untrusted or batch code-running jobs without touching the host.
Topics
FAQ
- What is SPACE?
- A sandbox platform that runs agent and user code in isolated Firecracker microVMs, controlled through an HTTP API. SPACE (Sandboxed Platform for Agentic Code Execution) gives applications a separate Linux environment for running code. Each sandbox is an instance of a template with its own virtual machine and writable filesystem, and several commands can run inside the same sandbox. Applications call the HTTP API to create sandboxes, start processes, read their output and exit status, and delete them when finished.
- Is SPACE free to use? What license does it have?
- SPACE is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is SPACE still maintained?
- Actively maintained. The last commit to SPACE was on Oct 18, 2026.
- What can I use SPACE for?
- Running model-generated code for an AI agent in an isolated environment. Providing each user a separate code workspace on a shared platform. Executing untrusted or batch code-running jobs without touching the host.
- Where can I get SPACE?
- The source code and installation instructions are on GitHub: https://github.com/perplexityai/SPACE
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A Linux CLI that embeds a Firecracker microVM (binary, kernel, initrd, rootfs) into a single executable that runs a container inside a VM.
A safe, private runtime that sandboxes autonomous AI agents, enforcing declared policies on files, system calls, and network access.
Desktop Kubernetes GUI built with Tauri and Rust that reads your kubeconfig and derives workload status the way kubectl does.
An all-in-one Docker sandbox giving AI agents a shared browser, shell, file system, VSCode Server, Jupyter and MCP services.
A Rust tool that intercepts dangerous shell commands and AI agent commands, requiring confirmation before risky operations run.