All-in-One Sandbox for AI Agents that combines Browser, Shell, File, MCP and VSCode Server in a single Docker container. It provides a unified, secure execution environment for AI agents and developers, streamlining workflows by integrating multiple functionalities and overcoming the limitations of single-purpose sandboxes.

How It Works

Built on cloud-native lightweight sandbox technology, AIO Sandbox consolidates browser, shell, file operations, VSCode Server, and MCP services within a single Docker container. Its core advantage is a unified filesystem, ensuring files accessed via the browser are immediately available in shell or file operations. It offers diverse interfaces like VNC, VSCode, Jupyter, and a terminal, all within a secure, zero-configuration environment ready for AI agent integration via MCP-compatible APIs.

Quick Start & Requirements

• Docker: Run docker run --rm -it -p 8080:8080 ghcr.io/agent-infra/sandbox:latest. A separate image is available for users in mainland China.
• SDKs: Install via pip install agent-sandbox (Python), npm install @agent-infra/sandbox (TypeScript/JavaScript), or go get github.com/agent-infra/sandbox-sdk-go (Golang).
• Prerequisites: Docker installation. SDKs require respective language environments.
• Access: The environment is accessible at http://localhost:8080/ for VNC Browser, VSCode Server, and MCP services. Documentation is available at http://localhost:8080/v1/docs.

Highlighted Details

• Unified Environment: Seamless workflows enabled by a shared filesystem across browser, VSCode, shell, and file operations.
• Browser Automation: Supports VNC for visual interaction, CDP for programmatic control, and MCP for high-level automation.
• Integrated Development Tools: Includes VSCode Server, Jupyter Notebook, WebSocket-based terminal, and port forwarding for web app previews.
• MCP Integration: Pre-configured Model Context Protocol servers for browser, file, shell, document processing (Markitdown), and research paper access (Arxiv).

Maintenance & Community

Developed by the Agent Infra team, the project welcomes contributions via its Contributing Guide. Community support is available through GitHub Discussions and an Issue Tracker.

Licensing & Compatibility

Released under the permissive Apache License 2.0, allowing for commercial use and integration into closed-source projects.

Limitations & Caveats

The README does not specify alpha status, known bugs, or unsupported platforms. Deployment configurations suggest recommended resource limits of 2GB memory and 1000m CPU per container.