Skip to content
AI agents and automation

Connect lobstr.io to Cursor and get web data from the editor

Add the lobstr.io MCP server to Cursor: its agent finds the operation, tells you its price, runs it and uses the records in your code.

With the lobstr.io MCP server, Cursor's agent gets web data while you code: it finds the right operation, tells you what it costs, runs it, and uses the records, to seed a database or write a test fixture.

Server: https://api.lobstr.io/v2/mcp

Before you start

  • A lobstr.io account. The free plan has 100 credits a month.
  • Your API key, from API & MCP in the app.
  • The Cursor desktop app.

Connect Cursor

  1. Open your MCP config

    Cursor reads its MCP servers from a JSON file:

    • ~/.cursor/mcp.json for every project on your machine
    • .cursor/mcp.json at a project's root for that project only

    Create the file if it doesn't exist.

  2. Add the lobstr.io server

    Paste this in, or add the lobstr entry to the mcpServers you already have, with your API key in place of YOUR_API_KEY:

    {
      "mcpServers": {
        "lobstr": {
          "url": "https://api.lobstr.io/v2/mcp",
          "headers": {
            "Authorization": "Bearer YOUR_API_KEY"
          }
        }
      }
    }

    In a project's file, keep the key out of version control: add .cursor/mcp.json to .gitignore.

  3. Check it works

    In Agent chat, ask "What's my lobstr.io balance?". The agent answers with your credits left and your plan.

What the agent can do

Tool What it does Credits
find_operations Finds the operations that fit what you ask, with their price Free
get_operation Reads an operation's options, the fields of its records, its errors Free
run_operation Runs an operation: one page of records, what it cost, the call for the next page The operation's price
get_balance Your credits left, and what a credit costs on your plan Free

By default Cursor asks before it uses a tool: click the arrow next to its name to see what it sends. You can let the free tools run without asking, and keep run_operation on approval to see every call that spends credits.

Ask in plain words

Get 20 restaurants in Lyon from Google Maps and save them
as fixtures/restaurants.json, with name, phone and rating.

The agent finds google_maps/places/search, runs it for 20 credits, and writes the file.

Building with the API itself? Point the agent at docs.lobstr.io/llms.txt: every page of the API docs, in Markdown.

Batches from Cursor

The batches (lists of inputs, schedules, exports) have their own server, https://mcp.lobstr.io/mcp, connected with your lobstr.io login (OAuth) instead of a key: add it as a second entry named lobstr-batches, "url": "https://mcp.lobstr.io/mcp" with no headers, and approve the connection in your browser.