This CLI tool allows you to build and manage E2B sandbox templates from your terminal.
You define your sandbox template in a Dockerfile
and then use the CLI to build
the sandbox template. You can then connect to the sandbox template via SDKs and
run your AI agents.
The Dockerfile
is the same as for Docker, but you can only use Debian based
linux distros as the base image.
npm install -g @e2b/cli
Then you can use the CLI like this:
e2b --help
- Authenticate with
e2b auth login
To authenticate without the ability to open browser, you can provide
E2B_ACCESS_TOKEN
as an environment variable. Get yourE2B_ACCESS_TOKEN
from e2b.dev/docs. Then use the CLI like this:E2B_ACCESS_TOKEN=sk_e2b_... e2b build
- Create a
Dockerfile
where you describe how your custom E2B sandbox template should look like. Majority of Debian based linux distros should work as the base image. Here is an example of a minimalDockerfile
for E2B sandbox template:
FROM ubuntu:22.04
-
Run
e2b build
inside the directory with theDockerfile
to create and build the sandbox template. You will get template ID that you use for connecting to the sandbox via SDKs and for rebuilding the sandbox template -
Use the template ID that you got during the
e2b build
with the Python or JS/TS SDK as theid
to create sandbox -
Rebuild the sandbox template by running
e2b build <id-of-the-template>
in the directory with theDockerfile
All commands can be called with a --path <path-to-dir>
flag that changes the
directory where the command will be called, without the need to call cd
.
-V, --version Display E2B CLI version
-h, --help display help for command