sandboxai-client 0.0.1

Secure sandboxes for AI agents and LLMs.

SandboxAI

Unleash your AI agents ... in a box!

SandboxAI is an open source runtime for executing AI-generated Python code and shell commands inside dynamic containers.

• Flexible control
  • Full-agency - Set the agent loose within the container
  • Limited-agency - Give the agent access to specific commands
• Container based
  • You define the environment
  • Integrate with existing tooling
• Pluggable backends
  • Run on your local machine or a remote VM
  • Run on your Kubernetes cluster (coming soon)
• Open REST API
  • Use the official Python or Go client
  • Generate your own client from the API spec

Where does it fit?

SandboxAI was designed to fit nicely into your AI software stack.

Quickstart

Make sure you have docker installed.

TIP: Free alternatives to docker desktop...

• On a mac? Checkout colima.
• On linux? Use docker engine.

docker ps

Install the python client.

pip install sandboxai-client

Run a sandbox on your local machine.

from sandboxai import Sandbox

with Sandbox(embedded=True) as box:
    print(box.run_ipython_cell("print('hi')"))
    print(box.run_shell_command("ls /"))

Contributing

We welcome contributions to SandboxAI. See the contributing doc for more information.

Contact

Let us know about features you are interested in seeing or reach out with questions. Visit our Discord channel to join the discussion!

Or just reach out on LinkedIn if you want to connect:

• Nick Stogner
• Sam Stoelinga