An all-in-one agent sandbox environment that combines Browser, Shell, File, MCP operations, and VSCode Server in a single Docker container, providing a unified, secure execution environment for AI agents and developers.
One-Minute Overview#
AIO Sandbox is an all-in-one agent sandbox environment that combines Browser, Shell, File, MCP operations, and VSCode Server in a single Docker container. It solves the problem of traditional single-purpose sandboxes by providing a unified execution environment where files are instantly available across different interfaces. It's the ideal solution for AI agents and developers who need seamless workflows without environment switching.
Core Value: One container integrates multiple development environments, eliminating barriers to agent development related to environment switching and file sharing.
Quick Start#
Installation Difficulty: Low - Deployable with a single Docker command
docker run --security-opt seccomp=unconfined --rm -it -p 8080:8080 ghcr.io/agent-infra/sandbox:latest
Is it suitable for my scenario?
- ✅ AI Agent Development: Provides a complete execution environment for AI with browser automation and code execution
- ✅ Multi-functional Testing: Ideal for scenarios requiring simultaneous testing of browser, terminal, and file operations
- ❌ High-performance Computing: Container resources are limited, not suitable for large-scale computing tasks
- ❌ Hardware-dependent Applications: Applications requiring specific hardware acceleration are not supported
Core Capabilities#
1. Unified File System - Solving File Isolation Issues#
- Seamless file sharing between browser, terminal, and file operations with instant availability Actual Value: AI agents don't need to handle file transfer and synchronization between different environments
2. Multi-Interface Integration - Providing Unified Development Experience#
- Integrates VNC browser, VSCode Server, Jupyter Notebook, and WebSocket terminal Actual Value: Developers and AI agents can complete all operations in one interface without switching tools
3. Secure Execution Environment - Ensuring Safe Code Execution#
- Provides sandboxed Python and Node.js execution environments with safety guarantees Actual Value: Safety assurance when executing untrusted code, suitable for AI agent code execution
4. Zero-Configuration MCP Services - Ready-to-Use Toolchain#
- Pre-configured MCP servers for browser automation, file operations, command execution, and document processing Actual Value: AI agents can directly use rich tools without complex configuration
Technology Stack & Integration#
Development Languages: Python, TypeScript/JavaScript, Golang Main Dependencies: Docker, Playwright, Chrome DevTools Protocol, VSCode Server, Jupyter Integration Method: API / SDK / Library
Maintenance Status#
- Development Activity: Actively maintained with regular new releases
- Recent Updates: Recent version updates show continuous project iteration
- Community Response: Active contributor community and issue feedback mechanism
Commercial & License#
License: Apache License 2.0
- ✅ Commercial Use: Allowed
- ✅ Modification: Allowed
- ⚠️ Restrictions: Must include copyright notice
Documentation & Learning Resources#
- Documentation Quality: Comprehensive
- Official Documentation: http://localhost:8080/v1/docs
- Example Code: Available in Python, TypeScript, and Golang