What can an AI agent do with E2B?

An E2B AI agent can create cloud sandbox environments, run code in multiple languages, execute shell commands, upload and read files, and tear down environments when done. This lets your AI agent safely execute real code and return actual output, not just write suggestions.

Is the code execution safe?

Yes. E2B runs all code in isolated cloud containers that are fully separated from your local machine and infrastructure. Each sandbox is ephemeral and disposed of after use, so there is no risk of side effects leaking between runs.

Which languages and runtimes does E2B support?

E2B supports Python, JavaScript, TypeScript, Bash, and other common runtimes out of the box. Custom sandbox templates let you pre-install additional dependencies or configure specific environments for your use case.

Do write actions like running code require my approval?

Yes. Any action that creates a sandbox, executes code, or modifies files is a write action and requires your explicit approval before the AI agent proceeds. Read-only actions like listing sandboxes or reading output files do not require approval.

Developer tools

E2B, run by an agent.

An AI agent that runs code, executes scripts, and manages secure cloud sandboxes.

Connect E2B Book a demo

What needs your approval

E2B never changes anything without you.

Anything that creates, edits or deletes in E2B waits for your approval. Approve a recurring task once and it runs unattended after that.

Connecting E2B

About 5 minutes

Auth: API key
Setup: About 5 minutes

Where to get access

Copy your API key from the E2B dashboard.

e2b.dev

Example tasks

Real work people hand to an agent that uses E2B.

generate a data analysis script and run it against an uploaded CSV in a secure sandbox
automatically test AI-generated code snippets and return pass/fail results to the user
run nightly regression scripts in isolated environments and post results to slack
execute user-submitted scripts safely without exposing your infrastructure