• READY
  • SCALED
  • CLEAN
  • HUMAN

How to provision and scale dev sandboxes with your AI

Connect Daytona to your AI client once. From then on, you manage the entire lifecycle of your ephemeral workspaces through chat.

Ask AI about this page

Short answer

How can I use my AI to manage dev sandboxes?

Your AI uses the Daytona MCP to trigger environment actions like creation, resizing, and deletion. You get full control over ephemeral workspaces without leaving your chat interface. You'll see how the AI handles the lifecycle of these environments below.

The outcomes

Where your environments land.

Every command you give your AI results in a specific state change in Daytona.

READY

Instant environments

The AI provisions a fresh sandbox for your current task. You get a ready-to-use workspace immediately.

SCALED

Adjusted resources

The AI modifies the hardware specs of an existing sandbox. This happens without you needing to touch a dashboard.

CLEAN

Resource recovery

The AI stops or deletes sandboxes when you're done. This keeps your cloud costs and clutter under control.

HUMAN

Manual oversight

You decide when a sandbox is no longer needed. The AI waits for your signal before destroying any work.

The workflow

What your AI does when the request arrives.

The AI translates your intent into specific Daytona tool calls.

  1. Provisioning

    When you need a new environment, the AI sets up a fresh sandbox with your specified configuration.

    create_sandbox

  2. Scaling

    If a task requires more power, the AI updates the sandbox resources to match the load.

    resize_sandbox

  3. Pausing

    To save costs without losing progress, the AI halts the running sandbox.

    stop_sandbox

  4. Cleanup

    Once the work is finished, the AI removes the sandbox entirely to free up resources.

    delete_sandbox

Try it

Copy these to start.

Use these prompts to trigger the Daytona tools.

Starting points

These are basic examples. Swap out the repo names or resource sizes to fit your specific project.

Daytona Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_AkqYHK4XFPqtgrdtnVnobFEfCwP5QQF4omfr3EXo/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — Daytona capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "daytona-dev-workspaces-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_AkqYHK4XFPqtgrdtnVnobFEfCwP5QQF4omfr3EXo/mcp"
    }
  }
}
Copy into chat04

  • Create a new sandbox for the 'frontend-app' repository.

  • Resize my current sandbox to use more CPU and memory.

  • Stop the sandbox I'm using for the 'api-service' testing.

  • Delete the sandbox associated with the 'legacy-migration' task.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect Daytona once, then ask.

Just link your account once. Your credentials stay encrypted, and you can start managing sandboxes in any MCP-compatible client immediately.

Connect Daytona to your AI

FAQ

How this task behaves

  • 01

    Can the AI delete my sandboxes without asking?

    No. The AI only executes the delete command when you explicitly tell it to remove a sandbox.

  • 02

    Can the AI change my cloud provider settings?

    No. The AI can only manage the sandboxes through the provided Daytona tools; it cannot modify your underlying cloud account or billing settings.

  • 03

    Will the AI automatically stop sandboxes to save money?

    No. The AI only stops a sandbox if you specifically request it or if you've set up a manual instruction to do so.

  • 04

    Can I resize a sandbox that has already been deleted?

    No. Once a sandbox is removed via the delete tool, it no longer exists to be resized.

  • 05

    Does the AI handle the actual code inside the sandbox?

    The AI manages the sandbox lifecycle, but it does not directly edit the files inside the environment unless you use a separate coding tool.

  • More questions about Daytona? The Connector page answers them. See everything the Daytona Connector can do

Connect Daytona to Claude, Cursor, ChatGPT & more