e2b

library2.46.0pypypi✓ verified 25d ago

E2B is an open-source infrastructure that provides isolated cloud sandboxes for AI agents to safely execute code, process data, and run tools. The Python SDK, currently at version 2.20.0, enables starting and managing these environments. Releases are frequent, often weekly or bi-weekly, addressing minor changes, patch fixes, and new features.

llm-agents devops aws gcp azure

pip install e2b

github ↗package ↗homepage ↗

INSTALL

IMPORT

SIG · E2B

e2b

llm-agentspythonv2.46.0

Install

4.9s avg

Import

1446ms

Disk

58MB

Pass rate

10/ 10

Env Coverage10 / 10

glibc

3.9–3.13

musl

3.9–3.13

Install & Compatibility

Where this runs

tested against v2.46.0 · pip install

no network on importno background threads

Install × environment matrix

Each cell = how many times install + import succeeded across repeated harness runs. Partial = flaky.
glibc = Debian/Ubuntu slim · musl = Alpine Linux

musl

py 3.10–3.910 runs

installs and imports cleanly · install 0.0s · import 1.550s · 62.6MB

glibc

py 3.10–3.910 runs

installs and imports cleanly · install 4.9s · import 1.342s · 63MB

58MB installed

● package 58MB

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

Sandbox

✓ from e2b import Sandbox

Use this for general sandbox management (commands, files, lifecycle).

Sandbox

✓ from e2b_code_interpreter import Sandbox

Use this specifically when you need `run_code()` functionality. Requires the `e2b-code-interpreter` package. In v2, the synchronous SDK's primary Sandbox class is imported from here.

This quickstart demonstrates how to create an E2B sandbox, run a shell command inside it, and print the output. It highlights the importance of setting the E2B_API_KEY environment variable for authentication.

import os from e2b import Sandbox # Ensure E2B_API_KEY is set in your environment # You can get your API key from the E2B dashboard: https://e2b.dev/docs/quickstart/running-your-first-sandbox api_key = os.environ.get('E2B_API_KEY', '') if not api_key: print("Warning: E2B_API_KEY environment variable not set. Sandbox creation may fail or default to a limited scope.") with Sandbox.create(api_key=api_key) as sandbox: print(f"Sandbox created with ID: {sandbox.sandbox_id}") result = sandbox.commands.run('echo "Hello from E2B!"') print(f"Stdout: {result.stdout}") if result.stderr: print(f"Stderr: {result.stderr}") print("Sandbox session ended.")

e2b --version

Debug

Known issues

breakingThe method for creating a Sandbox instance changed in v2. Instead of direct instantiation `Sandbox()`, you must now use the class method `Sandbox.create()`.

fix

Replace `sandbox = Sandbox()` with `sandbox = Sandbox.create()`.

affects: >=2.0.0

breakingFile operation APIs were updated in v2 for consistency. Single file writes now use `sandbox.files.write()`, and multiple file writes use `sandbox.files.write_files()`.

fix

Migrate file write operations from older patterns to `sandbox.files.write(path, content)` or `sandbox.files.write_files([{'path': ..., 'data': ...}])`.

affects: >=2.0.0

gotchaAn E2B API Key is required for most operations and must be provided, typically via the `E2B_API_KEY` environment variable or explicitly passed during sandbox creation. Failing to provide it will result in errors or limited functionality.

fix

Sign up on the E2B dashboard, obtain your API key, and set it as an environment variable (`export E2B_API_KEY=e2b_***`) or pass it as an argument (`Sandbox.create(api_key='e2b_***')`).

affects: All versions

gotchaThere are two distinct Python packages for E2B: `e2b` (for general sandbox management and commands) and `e2b-code-interpreter` (specifically for `run_code()` functionality). Both can export a `Sandbox` class, leading to potential import confusion if not installed and used correctly.

fix

Install `e2b-code-interpreter` separately if `run_code()` is needed. Pay attention to the import statements: `from e2b import Sandbox` for core SDK, and `from e2b_code_interpreter import Sandbox` for code interpretation.

affects: All versions

gotchaIn v2, sandboxes are secure by default. If you are using custom templates created before `envd v0.2.0`, you might need to rebuild them to enable secure communication, or you may encounter errors.

fix

Rebuild older custom templates or temporarily disable secure communication by setting `secure=False` during sandbox creation (not recommended for production). Check template `envd` version using `e2b template list` CLI command.

affects: >=2.0.0

Errors

Common errors & fixes

ImportError: cannot import name 'CodeInterpreter' from 'e2b_code_interpreter'

The `CodeInterpreter` class has likely been deprecated or moved, and the current recommended way to interact with the e2b sandbox for code execution is through the `Sandbox` class directly from the `e2b_code_interpreter` package or the main `e2b` package.

fix

Replace `from e2b_code_interpreter import CodeInterpreter` with `from e2b_code_interpreter import Sandbox` (if using the code interpreter specific package) or `from e2b import Sandbox` (for the general SDK), and use `Sandbox.create()` to initialize the sandbox.

e2b.exceptions.AuthenticationException: Authentication failed

This error occurs when the E2B API key is missing or invalid, preventing the SDK from authenticating with the E2B service.

fix

Ensure your E2B API key is correctly set as an environment variable named `E2B_API_KEY` or passed explicitly when creating the sandbox, for example: `sandbox = Sandbox.create(api_key='YOUR_API_KEY')`.

e2b.exceptions.TimeoutException: The sandbox is running but port is not open

This error often indicates that the sandbox timed out before the expected service (like a code interpreter) could become ready and open its port, or that the default start command is not properly configured for a custom template.

fix

Increase the sandbox timeout when creating it (e.g., `Sandbox.create(timeout=60)`), ensure your custom template is built on a `code-interpreter` base image, and verify the `setStartCmd` is correct if building from a Docker image.

RuntimeError: Attempted to access 'response.content' on a streaming response. Call 'response.read()' first.

This happens when trying to directly access the content of a streaming command response before it has been fully read and buffered, particularly with older SDK versions.

fix

Explicitly call the `.read()` or `.aread()` (for async) method on the command result object to ensure the entire output stream is consumed before attempting to access its content. For example: `result = sandbox.commands.run('ls -l'); output = result.read().stdout`

Upgrade

Version history

2.46.0latest on PyPI · released Aug 25, 2026

Audit

Dependencies

pythonrequiredRequired Python version.

e2b-code-interpreteroptionalRequired for `run_code()` functionality (e.g., executing arbitrary Python code within the sandbox).

Agent activity

51 hits · last 30 days

node

OpenAI (training)

Resources

githubhttps://github.com/e2b-dev/e2b ↗

packagehttps://pypi.org/project/e2b/ ↗

homepagehttps://e2b.dev/ ↗