All-in-One Sandbox for AI Agents that combines Browser, Shell, File, MCP and VSCode Server in a single Docker container.
AIO Sandbox packages a browser, shell, file system, VSCode Server, Jupyter Notebook, and MCP server into a single Docker container, giving AI agents a shared execution environment where files created in one tool are immediately accessible in another. The container exposes interfaces over a single port: a VNC remote desktop at `/vnc/`, a VSCode Server at `/code-server/`, and MCP-compatible API endpoints at `/mcp`, all reachable at `localhost:8080` after a one-line Docker run command. Agents interact with the environment through Python (`agent-sandbox`), TypeScript (`@agent-infra/sandbox`), or Go SDKs that wrap shell execution, file reads/writes, and browser screenshot calls. The browser layer supports VNC for visual interaction, Chrome DevTools Protocol for programmatic control, and MCP for higher-level automation. The project targets developers building Claude-based or other LLM-driven agents who need a reproducible, multi-capability sandbox without wiring together separate containers or managing cross-tool file sharing manually.
- ✓Open-source license (Apache-2.0)
- ✓Actively maintained (<30d)
- ✓Healthy fork ratio
- ✓Clear description
- ✓Topics declared
- ✓Documented (README)
git clone https://github.com/agent-infra/sandbox && cp sandbox/*.md ~/.claude/agents/Subagents overview
What people ask about sandbox
What is agent-infra/sandbox?
+
agent-infra/sandbox is subagents for the Claude AI ecosystem. All-in-One Sandbox for AI Agents that combines Browser, Shell, File, MCP and VSCode Server in a single Docker container. It has 5.1k GitHub stars and was last updated 14d ago.
How do I install sandbox?
+
You can install sandbox by cloning the repository (https://github.com/agent-infra/sandbox) or following the README instructions on GitHub. ClaudeWave also provides quick install blocks on this page.
Is agent-infra/sandbox safe to use?
+
Our security agent has analyzed agent-infra/sandbox and assigned a Trust Score of 100/100 (tier: Verified). See the full breakdown of passed checks and flags on this page.
Who maintains agent-infra/sandbox?
+
agent-infra/sandbox is maintained by agent-infra. The last recorded GitHub activity is from 14d ago, with 81 open issues.
Are there alternatives to sandbox?
+
Yes. On ClaudeWave you can browse similar subagents at /categories/agents, sorted by popularity or recent activity.
Deploy sandbox to your cloud
Ship this repo to production in minutes. Each platform spins up its own environment with editable env vars.
Maintain this repo? Add a badge to your README
Drop the badge into your GitHub README to show it's tracked on ClaudeWave. Each badge links back to this page and reflects the live Trust Score.
[](https://claudewave.com/repo/agent-infra-sandbox)<a href="https://claudewave.com/repo/agent-infra-sandbox"><img src="https://claudewave.com/api/badge/agent-infra-sandbox" alt="Featured on ClaudeWave: agent-infra/sandbox" width="320" height="64" /></a>More Subagents
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
The agent that grows with you
Java 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发
Production-ready platform for agentic workflow development.
The agent engineering platform.
🤯 LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations by hiring, scheduling, and reporting on your entire AI team.