Open-Source Sandbox SDK

3 weeks ago 2

Deep Research Agents

Computer Use Agents

Automations Agents

Background Agents

Reinforcement Learning

Secure MCPs

Deep Research Agents

Open-source, secure environment with real-world tools for enterprise-grade agents.

[.500]
[.873]
[.542]
[.704]
[.285]
[.717]
[.598]
[.557]
[.232]
[.746]

[.211]
[.013]
[.510]
[.718]
[.621]
[.223]
[.124]
[.801]
[.798]
[.117]️

[.817]
[.070]
[.353]

‍
[.833]
[.477]
[.620]

[.829]
[.195]
[.245]
[.891]
[.454]
[.145]
[.984]
[.634]
[.342]
[.746]

[.330]
[.103]
[.742]
[.004]
[.165]
[.459]
[.597]
[.910]
[.072]
[.336]

]·········[
]·········[
]·········[

RUNNING CODE…

[_______________]

[%%%%%__________]

[%%%%%%%%%%_____]

[%%%%%%%%%%%%%%%]

CPU: 8 × ▤ / RAM: 4 GB

]·········[
]·········[
]·········[

8 – ––––– ––– ––––– ––– ––––– –––
7 – ––––– ––– @@@@@ ––– ––––– –––
6 – ––––– ––– @@@@@ ––– ––––– –––
5 – @@@@@ ––– @@@@@ ––– ––––– –––
4 – @@@@@ ––– @@@@@ ––– ––––– –––
3 – @@@@@ ––– @@@@@ ––– @@@@@ –––
2 – @@@@@ ––– @@@@@ ––– @@@@@ –––
1 – @@@@@ ––– @@@@@ ––– @@@@@ –––
–––––––––––––––––––––––––––––––––
A B C

✓ CHART-1

______    ______    ______
❘    ❘_\  ❘    ❘_\  ❘    ❘_\
╔═══════╗ ╔═══════╗ ╔═══════╗
║ CSV ║ ║ TXT ║ ║ .JS ║
╚═══════╝ ╚═══════╝ ╚═══════╝
❘______❘  ❘______❘  ❘______❘

✓ File

╔ Email ══════════════╗
║ [email protected]      ║
╚═════════════════════╝
╔ Pw ═════════════════╗
║ ********            ║
╚═════════════════════╝
╔═════════════════════╗
║       Sign In       ║
╚═════════════════════╝

✓ UI

NVDA                            @
$120.91                        @
+32%                          @
                       @@@   @
            @@@       @   @ @
          @@   @     @     @
      @@@@      @   @
  @@@@           @@@
@
@

✓ CHART-2

1999 @@@@@@@@@@ │         │
1998 @@@        │         │
1997 @@@@@@@@@@@@@        │
1996 @@@@@@@    │         │
1995 @@@@@@@@@@@@@@@@     │
1994 @@@@@@@@@@@@@@@@@@@@@@@@@
1993 @@@@@      │         │
1992 @@@@@@@@@@ │         │
–––––––––––––––––––––––––––––––––
      2         4         6

✓ CHART-3

/!\
‍
Error: [$rootScope:inprog] $apply
already in progress

http://errors.angular.js.org/1.3
.15/$rootScope/inprog?p0=
%24apply

at angular.js:63

☓ Error

8 ––––│–––––––––│–––––––––│–––––@
7 ––––│–––––––––│–––@–––––│––––@–
6 ––––│–––––––––│––@–@–@@@@–––@––
5 ––––│–––––@@@–│–@–––@–––│@–@–@@
4 @@@@│–––@@–––@│@–––@–@––│–@–@––
3 ––––@@@@––@@@@@–––@–––@@│@@@–––
2 –––@│@@@@@––––│@@@––––––│––––––
1 –@@–│–––––––––│–––––––––│––––––
–––––––––––––––––––––––––––––––––
A B C

✓ CHART-4

88%

of Fortune 100 COMPANIES

Built for AI Agents, LLM Training, and MCPs

> HOVER (↓↓)

/EXPLORE

Deep research agents

Enable your agent to conduct time-consuming research on large datasets.

LEARN MORE

@@@
@@@@@@
@@@@@@@@@

AI data analysis & visualization

Connect your data to an isolated sandbox to securely explore data and generate charts.

LEARN MORE

======
========
=== ===
<

Coding agents

Securely execute code, use I/O, access the internet, or start terminal commands.

LEARN MORE

╔ ═ ╗
╣

╚ ═

Vibe coding

Use a sandbox as a code runtime for AI-generated apps. Supports any language and framework.

LEARN MORE

==╔═══╗==
==║ ✓ ║== ==╚═══╝==

Reinforcement learning

Use tens of thousands of concurrent sandboxes to run and evaluate reward functions.

LEARN MORE

Computer use

Use Desktop Sandbox to provide secure virtual computers in cloud for your LLM.

LEARN MORE

IN YOUR CODE
WITH A FEW LINES

1
2
3
4
5
6
7
8
9
10
11
12
~
~
~
~
~
~
‍~
~
~
~
~
~
~
~
~
~
~
~
~
~
~
~
~

// npm install @e2b/code-interpreter import { Sandbox } from '@e2b/code-interpreter' // Create a E2B Code Interpreter with JavaScript kernel const sandbox = await Sandbox.create() // Execute JavaScript cells await sandbox.runCode('x = 1') const execution = await sandbox.runCode('x+=1; x') // Outputs 2 console.log(execution.text)

1
2
3
4
5
6
7
8
9
10
~
~
~
~
~
~
‍~
~
~
~
~
~
~
~
~
~
~
~
~
~
~
~
~
~
~

# pip install e2b-code-interpreter from e2b_code_interpreter import Sandbox # Create a E2B Sandbox with Sandbox() as sandbox: # Run code sandbox.run_code("x = 1") execution = sandbox.run_code("x+=1; x") print(execution.text) # outputs 2

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35‍‍

// npm install ai @ai-sdk/openai zod @e2b/code-interpreter import { openai } from '@ai-sdk/openai' import { generateText } from 'ai' import z from 'zod' import { Sandbox } from '@e2b/code-interpreter' // Create OpenAI client const model = openai('gpt-4o') const prompt = "Calculate how many r's are in the word 'strawberry'" // Generate text with OpenAI const { text } = await generateText({ model, prompt, tools: { // Define a tool that runs code in a sandbox codeInterpreter: { description: 'Execute python code in a Jupyter notebook cell and return result', parameters: z.object({ code: z.string().describe('The python code to execute in a single cell'), }), execute: async ({ code }) => { // Create a sandbox, execute LLM-generated code, and return the result const sandbox = await Sandbox.create() const { text, results, logs, error } = await sandbox.runCode(code) return results }, }, }, // This is required to feed the tool call result back to the LLM maxSteps: 2 }) console.log(text)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28‍‍
~
~
~
~
~
~
~

# pip install openai e2b-code-interpreter from openai import OpenAI from e2b_code_interpreter import Sandbox # Create OpenAI client client = OpenAI() system = "You are a helpful assistant that can execute python code in a Jupyter notebook. Only respond with the code to be executed and nothing else. Strip backticks in code blocks." prompt = "Calculate how many r's are in the word 'strawberry'" # Send messages to OpenAI API response = client.chat.completions.create( model="gpt-4o", messages=[ {"role": "system", "content": system}, {"role": "user", "content": prompt} ] ) # Extract the code from the response code = response.choices[0].message.content # Execute code in E2B Sandbox if code: with Sandbox() as sandbox: execution = sandbox.run_code(code) result = execution.text print(result)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28‍‍
~
~
~
~
~
~
~

# pip install anthropic e2b-code-interpreter from anthropic import Anthropic from e2b_code_interpreter import Sandbox # Create Anthropic client anthropic = Anthropic() system_prompt = "You are a helpful assistant that can execute python code in a Jupyter notebook. Only respond with the code to be executed and nothing else. Strip backticks in code blocks." prompt = "Calculate how many r's are in the word 'strawberry'" # Send messages to Anthropic API response = anthropic.messages.create( model="claude-3-5-sonnet-20240620", max_tokens=1024, messages=[ {"role": "assistant", "content": system_prompt}, {"role": "user", "content": prompt} ] ) # Extract code from response code = response.content[0].text # Execute code in E2B Sandbox with Sandbox() as sandbox: execution = sandbox.run_code(code) result = execution.logs.stdout print(result)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30‍‍
~
~
~
~
~

# pip install mistralai e2b-code-interpreter import os from mistralai import Mistral from e2b_code_interpreter import Sandbox api_key = os.environ["MISTRAL_API_KEY"] # Create Mistral client client = Mistral(api_key=api_key) system_prompt = "You are a helpful assistant that can execute python code in a Jupyter notebook. Only respond with the code to be executed and nothing else. Strip backticks in code blocks." prompt = "Calculate how many r's are in the word 'strawberry'" # Send the prompt to the model response = client.chat.complete( model="codestral-latest", messages=[ {"role": "system", "content": system_prompt}, {"role": "user", "content": prompt} ] ) # Extract the code from the response code = response.choices[0].message.content # Execute code in E2B Sandbox with Sandbox() as sandbox: execution = sandbox.run_code(code) result = execution.text print(result)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25‍‍
~
~
~
~
~
~
~
~
~
~

# pip install ollama import ollama from e2b_code_interpreter import Sandbox # Send the prompt to the model response = ollama.chat(model="llama3.2", messages=[ { "role": "system", "content": "You are a helpful assistant that can execute python code in a Jupyter notebook. Only respond with the code to be executed and nothing else. Strip backticks in code blocks." }, { "role": "user", "content": "Calculate how many r's are in the word 'strawberry'" } ]) # Extract the code from the response code = response['message']['content'] # Execute code in E2B Sandbox with Sandbox() as sandbox: execution = sandbox.run_code(code) result = execution.logs.stdout print(result)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30‍‍
~
~
~
~
~

# pip install langchain langchain-openai e2b-code-interpreter from langchain_openai import ChatOpenAI from langchain_core.prompts import ChatPromptTemplate from langchain_core.output_parsers import StrOutputParser from e2b_code_interpreter import Sandbox system_prompt = "You are a helpful assistant that can execute python code in a Jupyter notebook. Only respond with the code to be executed and nothing else. Strip backticks in code blocks." prompt = "Calculate how many r's are in the word 'strawberry'" # Create LangChain components llm = ChatOpenAI(model="gpt-4o") prompt_template = ChatPromptTemplate.from_messages([ ("system", system_prompt), ("human", "{input}") ]) output_parser = StrOutputParser() # Create the chain chain = prompt_template | llm | output_parser # Run the chain code = chain.invoke({"input": prompt}) # Execute code in E2B Sandbox with Sandbox() as sandbox: execution = sandbox.run_code(code) result = execution.text print(result)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
~
~
~
~
~
~
~
~
~
~
~
~

from llama_index.core.tools import FunctionTool from llama_index.llms.openai import OpenAI from llama_index.core.agent import ReActAgent from e2b_code_interpreter import Sandbox # Define the tool def execute_python(code: str): with Sandbox() as sandbox: execution = sandbox.run_code(code) return execution.text e2b_interpreter_tool = FunctionTool.from_defaults( name="execute_python", description="Execute python code in a Jupyter notebook cell and return result", fn=execute_python ) # Initialize LLM llm = OpenAI(model="gpt-4o") # Initialize ReAct agent agent = ReActAgent.from_tools([e2b_interpreter_tool], llm=llm, verbose=True) agent.chat("Calculate how many r's are in the word 'strawberry'")

Features your
agents will love

AI agents need real-world tools to complete superhuman level tasks.

> MADE FOR AI

> DSCVR ALL (↓↓)

Works with any LLM

Use OpenAI, Llama, Anthropic, Mistral, or your
own custom models. E2B is LLM-agnostic
and compatible with any model.

Quick start

The E2B Sandboxes in the same region as
the client start in less than 200 ms.

NO COLD STARTS

Run

... or just any other AI-generated code.

AI-generated Python, JavaScript, Ruby, or C++? Popular framework or custom library? If you can run it on a Linux box, you can run it in the E2B sandbox.

Secure quick start

The E2B Sandboxes in the same region as
the client start in less than 200 ms.

NO COLD STARTS

Features made for agents

The full-stack of secure tools for any agentic workflow.

COMPUTERS FOR AGENTS

Secure & isolated

Each sandbox is powered by Firecracker, a microVM made to run untrusted workflows.

FULL ISOLATION

Up to 24h long sessions

Run for a few seconds or several hours, each E2B
sandbox can run up to 24 hours.

AVAILABLE IN PRO

Install any package or system library with

and more.

Completely customize the sandbox for your use case by creating a custom sandbox template or installing a package when the sandbox is running.

BYOC, on-prem, or self-hosted

E2B works anywhere you do.
In your AWS, GCP, or Azure account, or your VPC.

GET INSPIRed BY
OUR COOKBOOK

Production use cases & full-fledged apps.

HOVER (↓↓)

“E2B runtime unlocked rich answers and interactive charts for our users. We love working with both the product and team behind it.”
— Denis Yarats, CTO

“E2B allows us to scale-out training runs by launching hundreds of sandboxes in our experiments, which was essential in Open R1.”
— Lewis Tunstall, Research Engineer

“It took just one hour to integrate E2B end-to-end. The performance is excellent, and the support is on another level. Issues are resolved in minutes.”
— Maciej Donajski, CTO

“E2B has revolutionized our agents' capabilities. This advanced alternative to OpenAI's Code Interpreter helps us focus on our unique product.”
— Kevin J. Scott, CTO/CIO

“Manus doesn’t just run some pieces of code. It uses 27 different tools, and it needs E2B to have a full virtual computer to work as a real human.”
— Tao Zhang, Co-founder

“Executing Athena’s code inside the sandbox makes it easy to check and automatically fix errors. E2B helps us gain enterprises’ trust.”
— Brendon Geils, CEO

“We needed a fast, secure, and scalable way for code execution. E2B’s API interface made their infrastructure almost effortless to integrate.”
— Benjamin Klieger, Compound AI Lead

“We implemented E2B in a week, needing just one engineer working in spare cycles. Building it in-house would've taken weeks and multiple people.”
— Luiz Scheidegger, Head of Engineering

“LLM-generated API integrations to external services make Gumloop incredibly useful. E2B is essential to make that happen at scale and to deliver reliable performance to our users.”
— Max Brodeur-Urbas, CEO

Open-source, secure environment with real-world tools for enterprise-grade agents.

/RUN CODE

>
>
‍
>>

Read Entire Article

Open-Source Sandbox SDK