mind code
Browse documentation

Sandbox

Isolated execution environment for the Mind Web agent

The sandbox provides a secure, isolated environment where the Mind Web agent executes tasks. Each task runs in its own sandbox with configurable access controls, ensuring your code and resources remain protected.

How it works

When you assign a task, the agent:

  1. Spins up an isolated sandbox environment
  2. Clones authorized repositories into the sandbox
  3. Configures the environment based on detected project settings
  4. Executes the task with access only to explicitly permitted resources
  5. Tears down the sandbox when the task completes

Browser automation

Sandboxes can run a headless Chromium browser in the same isolated environment so the agent can open pages it serves locally — for example validating UI changes or reproducing a browser-based bug. Dedicated browser MCP servers (Playwright MCP, Chrome DevTools MCP) are not preinstalled; configure a custom MCP server or Agent Plugin when your workflow needs a specific browser stack.

Ask the agent to verify a page in the sandbox when Chromium is available in that environment; otherwise bring your own browser tooling.

Configuration

  • Internet Access — Control which domains the agent can reach
  • Agent Plugins and MCP — Install Agent Plugins or configure custom MCP servers
  • Environment Variables — Configure environment variables and secrets for task execution
  • Environment Configuration — Configure the sandbox environment for your project