# E2B — Secure Sandboxes for AI Code

> E2B runs AI-generated code in isolated cloud sandboxes. Install the Python/JS SDK, set `E2B_API_KEY`, then execute commands safely inside a sandbox.

## Install

Save as a script file and run:

# E2B — Secure Sandboxes for AI Code

> E2B runs AI-generated code in isolated cloud sandboxes. Install the Python/JS SDK, set `E2B_API_KEY`, then execute commands safely inside a sandbox.

## Quick Use

1. Install:
   ```bash
   pip install e2b
   ```
2. Run:
   ```bash
   python -c "from e2b import Sandbox; print('e2b import ok')"
   ```
3. Verify:
   - Create a sandbox and run `echo "Hello from E2B!"` using the README example


---

## Intro

E2B runs AI-generated code in isolated cloud sandboxes. Install the Python/JS SDK, set `E2B_API_KEY`, then execute commands safely inside a sandbox.

- **Best for:** agent builders who need to execute untrusted code with isolation (CI-like sandboxes) instead of running directly on the host machine
- **Works with:** Python SDK (`pip install e2b`) or JS SDK (`npm i e2b`), E2B API key, sandbox command execution (repo examples)
- **Setup time:** 12 minutes


### Quantitative Notes

- Install: `pip install e2b` (repo)
- Setup time ~12 minutes
- GitHub stars (verified): see Source & Thanks


---

## Practical Notes

Use E2B to separate *reasoning* from *execution*. Let your LLM plan the steps, but run risky commands in a sandbox with clean state. This is especially useful for agents that generate code, install packages, or process untrusted data. In production, standardize images/templates and capture logs + artifacts for auditability.

**Safety note:** Isolation reduces risk, but you still need secret management—never bake API keys into prompts or artifacts.

### FAQ

**Q: Is E2B self-hostable?**
A: The README links to a self-hosting guide and infrastructure repo for running E2B on your own cloud.

**Q: What can I run inside a sandbox?**
A: Commands and code execution via SDK methods; start with shell commands, then add a code-interpreter SDK if needed (repo mentions).

**Q: How do I control cost?**
A: Reuse sandboxes when possible, keep jobs short, and scope execution to minimal steps per task.

---

## Source & Thanks

> GitHub: https://github.com/e2b-dev/E2B
> Owner avatar: https://avatars.githubusercontent.com/u/129434473?v=4
> License (SPDX): Apache-2.0
> GitHub stars (verified via `api.github.com/repos/e2b-dev/E2B`): 12,146


---

<!-- ZH -->

# E2B——为 AI 生成代码提供安全沙盒

> E2B 在云端隔离沙盒里运行 AI 生成代码。安装 Python/JS SDK 并设置 `E2B_API_KEY`，即可在 sandbox 内安全执行命令与运行代码，显著降低误操作与泄露风险。

## 快速使用

1. 安装：
   ```bash
   pip install e2b
   ```
2. 运行：
   ```bash
   python -c "from e2b import Sandbox; print('e2b import ok')"
   ```
3. 验证：
   - Create a sandbox and run `echo "Hello from E2B!"` using the README example


---

## 简介

E2B 在云端隔离沙盒里运行 AI 生成代码。安装 Python/JS SDK 并设置 `E2B_API_KEY`，即可在 sandbox 内安全执行命令与运行代码，显著降低误操作与泄露风险。

- **适合谁（Best for）:** 要执行不可信代码的 agent 开发者，希望用隔离沙盒（类似 CI 环境）替代直接在宿主机运行
- **兼容工具（Works with）:** Python SDK（`pip install e2b`）或 JS SDK（`npm i e2b`）、E2B API key、在 sandbox 内执行命令（仓库示例）
- **安装时间（Setup time）:** 12 分钟


### 量化信息

- 安装：`pip install e2b`（仓库）
- 装机约 12 分钟
- GitHub stars（已核验）：见「来源与感谢」


---

## 实战要点

把 E2B 用在“思考”和“执行”的隔离上：LLM 负责计划，但高风险命令在干净的 sandbox 里跑。对会生成代码、装包或处理不可信数据的 agent 特别有用。上生产后建议标准化镜像/模板，并保留日志与产物，便于审计与复现。

**安全提示：** 隔离能降风险，但仍要做好密钥管理——不要把 API key 写进 prompt 或产物里。

### FAQ

**Q: E2B 能自建吗？**
A: README 提供了 self-hosting 指南与基础设施仓库链接，可在你自己的云上部署。

**Q: sandbox 里能跑什么？**
A: 可以通过 SDK 在 sandbox 内运行命令与代码；先从 shell 命令开始，必要时再接入 code-interpreter SDK（仓库提到）。

**Q: 怎么控制成本？**
A: 尽量复用 sandbox、把任务拆小并缩短运行时间，每次只做必要步骤。

---

## 来源与感谢

> GitHub：https://github.com/e2b-dev/E2B
> Owner avatar：https://avatars.githubusercontent.com/u/129434473?v=4
> 许可证（SPDX）：Apache-2.0
> GitHub stars（已通过 `api.github.com/repos/e2b-dev/E2B` 核验）：12,146


---
Source: https://tokrepo.com/en/workflows/e2b-secure-sandboxes-for-ai-code
Author: Agent Toolkit