Deploy straight from Cursor

SimDeploy lets Cursor deploy a project with one non-interactive command through a native MCP server — the agent analyzes, deploys, and reads logs without leaving the editor.

Add the SimDeploy MCP server to Cursor and the agent gains deploys, cost analysis and log reading — without leaving the editor and with zero infrastructure decisions.

How it works

1. Add {"mcpServers": {"simdeploy": {"command": "simdeploy-mcp"}}} to .cursor/mcp.json
2. simdeploy login  (reads the token securely — once, by a human)
3. Ask the agent: deploy this project
4. The agent gets the public URL or the typed error with logs

—MCP analysis tools work offline (nothing leaves the machine)
—deploy_project builds locally and uploads only the output
—Estimated cost per architecture before publishing

How Cursor deploys through MCP

Cursor gains deploys by adding SimDeploy's MCP server to .cursor/mcp.json — the simdeploy-mcp binary ships with the CLI. Once a human authenticates with simdeploy login, the agent can analyze the project, estimate cost, deploy and read logs without leaving the editor. The analysis tools run locally, so framework detection, route discovery and the cost projection never send your code off the machine; only the built output is uploaded when the agent deploys. On failure the agent receives a typed error with per-stage logs instead of a dead end.

What deploys today

Static builds deploy end-to-end today, including static exports. Server-rendered and serverless targets are planned and surface as a typed error rather than a partial deploy. Because the token lives in the environment after simdeploy login, it is never pasted into a Cursor chat, and it stays scoped and revocable.

Deploy your first project Agent docs

FAQ

How do I deploy from Cursor?

Add the SimDeploy MCP server to your project, authenticate once with simdeploy login, then ask the agent to deploy this project — it returns the public URL or a typed error with logs.

Where do I configure the MCP server in Cursor?

In .cursor/mcp.json, add {"mcpServers": {"simdeploy": {"command": "simdeploy-mcp"}}}. The simdeploy-mcp binary ships with the CLI.

Does analysis run offline in Cursor?

Yes. The MCP analysis tools run locally — nothing leaves the machine. Only the built output is uploaded when the agent deploys.

What can I deploy from Cursor today?

Static builds ship today. Server-rendered and serverless targets are planned.