Skip to main content

​
Overview

Connect Operator to Cursor so its AI assistant can reach beyond local context and run targeted, adaptive multi-source searches whenever it hits a confusing error, missing docs, or unfamiliar library. Instead of guessing from partial training data, the agent can call Operator to pull in current documentation, GitHub issues, blog posts, and changelogs. This page shows how to wire Operator in as an MCP server for Cursor, both at the project level and globally, and suggests a few prompts you can use to make sure the integration is doing real work for your team.

​
Why use Operator inside Cursor?

  • Keep Cursor focused on editing your code while Operator handles heavy-duty research.
  • Get answers that reference specific upstream docs, issues, and examples instead of vague suggestions.
  • Stay current with fast-moving libraries without constantly leaving the editor for manual search.
  • Share a single Operator API key across multiple projects and tools.
Under the hood, Cursor treats Operator as a tool exposed via MCP. You do not have to manage tool invocation directly. Just ask Cursor to “use Operator” when it needs deeper context.

​
Configure Operator as a project-level MCP server

To enable Operator for a single project, create a .cursor/mcp.json file in your repo and add the Operator MCP server: 
{
  "mcpServers": {
    "operator": {
      "url": "https://mcp.operator.io/mcp",
      "headers": {
        "Authorization": "Bearer ${OPERATOR_API_KEY}"
      }
    }
  }
}
Set OPERATOR_API_KEY in your environment before launching Cursor, for example: 
export OPERATOR_API_KEY="sk_your_api_key_here"
You can create and manage your key at operator.io. If you prefer SSE, change the url to https://mcp.operator.io/sse. You can also send the API key with an x-api-key header instead of Authorization. Project-level configuration is great when you want fine-grained control or when only a subset of repos should use Operator.

​
Global setup for all Cursor projects

To make Operator available in every Cursor workspace, configure a global MCP file in your home directory: 
~/.cursor/mcp.json

{
  "mcpServers": {
    "operator": {
      "url": "https://mcp.operator.io/mcp",
      "headers": {
        "Authorization": "Bearer ${OPERATOR_API_KEY}"
      }
    }
  }
}
Global settings ensure new projects automatically have access to Operator-powered search without extra configuration. As with project-level config, you can switch to SSE by changing the url to https://mcp.operator.io/sse. You can always override or disable Operator for a specific repository by adjusting its local .cursor/mcp.json if needed.

​
Using Operator inside Cursor

Once configured, Cursor’s agent can call Operator as a tool whenever it needs external context. Typical use cases include:
  • Looking up fresh documentation for rapidly evolving libraries.
  • Finding GitHub issues that match error messages from your build or tests.
  • Pulling migration guides when you upgrade frameworks like Next.js or Prisma.
  • Gathering real-world examples for unfamiliar integrations.
When you want to be explicit, you can say things like:
  • “Use Operator to find current docs for this TypeScript error.”
  • “Ask Operator for migration notes from Next.js 14 to 15.”

​
Troubleshooting and verification

If Cursor does not appear to call Operator, double-check that your MCP configuration file is valid JSON and that your API key is set correctly. Configuration errors are often surfaced in Cursor’s logs or output panel. For more details on using Operator with MCP clients, see the Operator quickstart and integration docs at docs.operator.io.