Documentation

This guide helps you migrate from E2B SDK v1 to v2, covering all breaking changes and new patterns.

Table of contents

SDK v2 introduces several important changes:

Breaking changes

1. Sandbox creation in synchronous Python SDK

In v2, the synchronous Python SDK uses a class method create() instead of the constructor Sandbox().

from e2b_code_interpreter import Sandbox

sandbox = Sandbox.create()
sandbox = Sandbox.create(template="base")

In the v1, you would instantiate directly:

from e2b_code_interpreter import Sandbox

sandbox = Sandbox()
sandbox = Sandbox(template="base")

2. Secure communication by default

Sandboxes are now secure by default. That means you cannot access the sandbox directly via its URL; there's authentication required. If you use the SDK, this is handled for you automatically.

This shouldn't affect you, but if you need to access sandbox directly without SDK and you are now getting 401, you may need to set secure to false.

If you are using custom code interpreter templates, it may be necessary to rebuild the template to ensure compatibility with secure sandboxes.

import { Sandbox } from '@e2b/code-interpreter'

const sandbox = await Sandbox.create({ secure: false }) // Explicitly disable

3. File writing in Python SDK

The file writing API in Python has been made more consistent.

In v2, use sandbox.files.write() for single files and sandbox.files.write_files() for multiple files:

from e2b_code_interpreter import Sandbox

sandbox = Sandbox.create()

# Write single file
info = sandbox.files.write("/tmp/file.txt", "content")

# Write multiple files
infos = sandbox.files.write_files([
    {"path": "/tmp/file1.txt", "data": "content1"},
    {"path": "/tmp/file2.txt", "data": "content2"}
])

In v1, the same write() method was overloaded for both single and multiple files:

from e2b_code_interpreter import Sandbox

sandbox = Sandbox.create()

# Write single file
info = sandbox.write(path="/tmp/file.txt", data="content")

# Write multiple files
infos = sandbox.write([
    {"path": "/tmp/file1.txt", "data": "content1"},
    {"path": "/tmp/file2.txt", "data": "content2"}
])

4. Listing sandboxes

The method for listing sandboxes has been updated to use pagination.

import { Sandbox, SandboxInfo } from '@e2b/code-interpreter'

// Get paginator
const paginator = Sandbox.list()

// Iterate through all sandboxes
for (const sandbox of await paginator.nextItems()) {
    console.log(sandbox.sandboxId)
}

// Iterate through all sandboxes
const allSandboxes: SandboxInfo[] = []
while (paginator.hasNext) {
    const items = await paginator.nextItems()
    allSandboxes.push(...items)
}

// With query
const queryPaginator = Sandbox.list({query: {metadata: {key: "value"}}})