# agent-infra/sandbox

All-in-One Sandbox for AI Agents that combines Browser, Shell, File, MCP and VSCode Server in a single Docker container.

Repository: https://github.com/agent-infra/sandbox
Canonical: https://ross.abutalabs.com/products/sandbox
Homepage: https://sandbox.agent-infra.com
Language: Python
License: Apache-2.0
License Family: permissive
Topics: agent, all-in-one, sandbox, browser, filesystem, mcp, shell
Last push: 2026-08-25T08:10:03+00:00

## Health v2 (maintenance only)
Score: 82/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 90, longevity 28
- inputs: {"age_days": 392, "days_push": 8, "days_rel": 71, "gap_med": 3, "n_releases_24m": 20}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5790, forks 515 (observed 2026-08-28T04:09:29.595435+00:00)

## What it is
AIO Sandbox is an all-in-one Docker container environment for AI agents that bundles a browser (with VNC), shell terminal, filesystem, VSCode Server, Jupyter, and an MCP server behind a single API/SDK. It provides isolated, secure code execution and browser automation for agent workflows.

## Use cases
- run a sandboxed environment for AI agents to execute code and browse the web
- give an LLM agent shell, file, and browser access in one container
- self-host a code execution sandbox for computer-use agents
- run browser automation with screenshots via an API
- host an MCP server exposing browser, terminal, and file tools
- provide cloud VSCode and Jupyter for agent-driven development

## When to choose
- you need a single container combining browser, shell, file, and MCP tooling for agents
- you want safe, isolated Python/Node.js code execution for LLM agents
- you want one unified API/SDK instead of wiring multiple services
- you need cloud VSCode, Jupyter, and VNC browser access out of the box

## When to avoid
- you only need simple shell command execution without browser or IDE features
- you require strong multi-tenant security isolation beyond a single Docker container
- you cannot run Docker or containers in your environment
- you need a lightweight sandbox with minimal resource footprint

## Facets
- artifact type: service
- maturity: active
- function: agent-framework, mcp, http-server, web-scraping, file-system, developer-tools
- domain: developer-tools, self-hosted
- platform: self-hosted, python, cloud
- tags: sandbox, browser-automation, code-execution, vscode-server, jupyter, vnc, all-in-one, agent-environment, ai-agents, automation, docker, nodejs

## Member repositories
- agent-infra/sandbox (main) score 82

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:29.595435+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:52:48.786916+00:00, confidence not recorded.
  - readme: https://github.com/agent-infra/sandbox (fetched 2026-08-28T04:09:29.595435+00:00, sha e7446c38c641)
  - homepage: https://sandbox.agent-infra.com (fetched 2026-08-29T08:48:02.845881+00:00, sha c6c2ee63e3c8)
- Data as of 2026-08-30T08:39:29.467469+00:00.