Find tools › Open-source projects › Write code faster

box

Open-source project · by strands-agents

A secure container that stops AI agents from running harmful commands, stealing secrets, or reading private files.

Some setup

What it costs

Free and open source; you may pay for the AI model it uses.

Use it to

  • Isolate coding agents from sensitive host files and networks
  • Inject credentials into API requests without exposing raw keys
  • Enforce fine-grained allow and deny policies on tool calls

How to start

  1. Download the binaries to a local folder: curl -fsSL https://raw.githubusercontent.com/strands-agents/box/main/download.sh | sh

  2. Verify the installation by running ./box-core/box --version in your terminal.

  3. Define environment grants in box.toml and fine-grained rule policies in policy.dw.

  4. Run your chosen agent application inside the sandboxed environment.

Good to know. Currently in preview and only supports macOS 15 or later on Apple silicon; Linux support is planned but not yet released.

Before you install

  • Found automaticallyNot reviewed by a person yet. Check it before installing.
  • License: Apache-2.0You are allowed to use it under these terms.
  • Last updated 2026-11-18

More detail

An open-source sandbox engine that restricts what AI agents can execute, read, write, and access over the network.

Autonomous agents can execute dangerous shell commands, leak credentials, or access sensitive files. Reach for Box when you want OS-level isolation and default-deny policies around an agent without exposing host secrets.

Similar tools

Descriptions are written in plain English from each project's own documentation and refreshed by our agent; this one was last checked 2026-10-09. Spotted a mistake? · Open in Find tools · © 2026 Which AI Map