You need a desktop or browser to run an agent, but managing VMs and images is slow and error-prone. The /v1/machines API lets you provision cloud machines on demand. You start, stop, and snapshot environments that your computer use agent can control. This solves the setup friction for any task that needs a live UI.
How it works
POST /v1/machines creates a cloud machine with a machine_id. The API returns an environment you can start, stop, or snapshot. You control the state from your code without managing VMs directly. After provisioning, you use that machine_id when you start a computer use agent run or workflow.
curl -s https://coasty.ai/v1/machines \
-H "X-API-Key: $COASTY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"machine_type": "browser",
"os": "linux",
"size": "small"
}' | jq .Key fields
- machine_type: browser, desktop, or terminal
- os: linux, windows, or macos
- size: small, medium, or large
- machine_id: returned by the API to identify the environment
- Actions: start, stop, and snapshot via machine_id
POST /v1/machines to get a machine_id, then start and control the environment from your agent runs.
Where this beats brittle automation
Traditional automation relies on selectors, XPath, or hardcoded IDs that break when UI changes. A computer use agent sees the screen and clicks like a human. With a real machine, the agent interacts with actual apps, browsers, and terminals. You avoid fragile selectors and get robust behavior across updates.
Next steps
Provision a machine, then start a run or workflow on it. Use the /v1/runs endpoint with your machine_id to drive the agent. Check the Coasty docs for workflows and the MCP server if you want to control Coasty from Cursor or Claude Desktop. Get a key at https://coasty.ai/developers to begin.
Want to see this in action?
View Case Studies