ScrapeField in Cursor
One entry in mcp.json gives Cursor’s agent every endpoint as a tool. Without a key it answers in demo mode; with one, it reads live data.
Add it
{
"mcpServers": {
"scrapefield": { "url": "https://scrapefield.com/mcp" }
}
}The file is .cursor/mcp.json in a project, or ~/.cursor/mcp.json for every project. ${env:SCRAPEFIELD_API_KEY} reads the key from your environment, so it never sits in the file.
From Cursor’s own documentation, read on 9 October 2026. A key comes from signing up; an agent can also ask for one for you with sign_up (how).
Things to ask it
In plain words. The agent picks the tool; beside each is the one it reaches for, and what a call costs.
| Ask | Tool | Credits |
|---|---|---|
| Find the best-rated coffee shops in Austin on Google Maps. | search_places | 3 |
| What do the newest reviews of that café complain about? | get_place_reviews | 6 |
| Which companies are hiring backend engineers in Berlin this week? | search_linkedin_jobs | 3 |
| What has Ferrari posted on X lately, and which post got the most views? | get_x_posts | 3 |
| Find TikTok videos about latte art, most viewed first. | search_tiktok | 3 |
With no key
Every tool answers in demo mode: one real answer per tool, captured once a week, in the live shape, whatever you ask, charged nothing. It shows what each tool returns before anyone signs up. Each platform’s tools, with their prices: Google Maps, LinkedIn, Instagram, TikTok and X. Every tool, with its input schema, is in the MCP server’s docs.