langchain-e2b

Quick Install

pip install langchain-e2b

Deep Agents SDK

from e2b import Sandbox
from langchain_e2b import E2BSandbox

e2b_sandbox = Sandbox.create()
backend = E2BSandbox(sandbox=e2b_sandbox)

try:
    result = backend.execute("echo hello")
    print(result.output)
finally:
    e2b_sandbox.kill()

For async agents, use E2B's async SDK with the async backend:

from e2b import AsyncSandbox
from langchain_e2b import AsyncE2BSandbox

e2b_sandbox = await AsyncSandbox.create()
backend = AsyncE2BSandbox(sandbox=e2b_sandbox)

try:
    result = await backend.aexecute("echo hello")
    print(result.output)
finally:
    await e2b_sandbox.kill()

Deep Agents Code

Install langchain-e2b into the dcode environment, then run with the E2B sandbox provider:

dcode --install langchain-e2b --package
export E2B_API_KEY=...
dcode --sandbox e2b

What is this?

langchain-e2b adapts an existing E2B sandbox to the Deep Agents sandbox protocol. It uses the low-level e2b SDK so Deep Agents can run shell commands and move files through the standard Deep Agents sandbox interface.

For SDK use, this package intentionally does not hide E2B sandbox lifecycle management. Use the E2B SDK to create, connect to, configure, and kill sandboxes, then pass the connected sandbox object to E2BSandbox.

For Deep Agents Code, the package also exposes a dcode sandbox provider entry point. The provider creates, reconnects to, and deletes E2B sandboxes for dcode --sandbox e2b.

Contributing

Contributions are welcome. Keep the adapter focused on implementing the Deep Agents sandbox backend protocol over the official E2B SDK.

Development

uv sync --group test
make test
make lint

Integration tests require E2B_API_KEY:

E2B_API_KEY=... make integration_tests