Ask your AI for a scraper. Get an app.
Tell Claude, Cursor, Replit or Lovable "create a Google rank tracker using UnblockingAPI". It reads our build guide, picks the right templates and writes an app that fetches the sites it needs - past bot walls, from any country, as JSON where it can.
Create a Google rank tracker using UnblockingAPI. I add keywords and my domain, it checks rankings daily from the US and shows each keyword's current position, the change since yesterday and its history. Read UnblockingAPI's build guide first (the get_build_guide tool, or https://github.com/unblockingapi/mcp/tree/main/guides) and call its REST API directly from the backend with plain fetch, my API key stored as a secret.
How it works
Your agent writes the app; UnblockingAPI is the part that gets the pages.
Connect UnblockingAPI to your AI
Add the MCP server to Claude, Cursor, Replit, Lovable or any MCP app. It brings a build guide your agent reads before it writes code.
Ask for the app you want
"Create a Google rank tracker using UnblockingAPI." Your agent plans the data model, the scheduled job and the UI.
It calls the API directly
The app fetches pages with plain HTTP from its backend, your key stored as a secret. JSON where a template exists, HTML where not.
Connect once: https://mcp.unblockingapi.com/mcp - setup guides for every app.
Recipes
A prompt to paste, what your agent will build, and what it costs to run.
Google rank tracker
Keywords, domains and countries in; daily positions and history out.
1 credit per keyword per check: 100 keywords checked daily is about 3,000 credits a month.
Open the recipeReal estate listings aggregator
Several agencies' listings in one searchable page, refreshed on a schedule.
1 credit per results page per refresh: 5 sources x 3 pages x 4 runs a day is about 1,800 credits a month.
Open the recipePrice monitor
Paste product URLs and a target; get alerted when the price moves.
1 credit per product per check: 50 products twice a day is about 3,000 credits a month.
Open the recipeAny other app
A job board aggregator, a competitor watcher, a lead list from a directory: describe it and add this. The guide covers the API, keeping the key secret, getting complete pages and templates.
Build <your app> using UnblockingAPI to fetch the websites it needs.
Read UnblockingAPI's build guide first (the get_build_guide tool, or
https://github.com/unblockingapi/mcp/tree/main/guides) and call its REST API directly from the backend
with plain fetch, my API key stored as a secret.Questions
No, but it helps. Connected, your agent can call get_build_guide and look up templates with their exact output shapes, and test fetches before it writes code. Without it, point the agent at the build guides on GitHub and the API docs.
Whatever your project uses. The API is plain HTTPS with JSON, so the agent calls it with the language's own HTTP client - fetch in JavaScript and TypeScript, the standard library in Python - and needs no SDK.
Into the platform's secret store, used only from server code: a Supabase Edge Function secret in Lovable, Replit Secrets, environment variables in Bolt, Vercel or Next.js. The build guide tells the agent to never put it in browser code.
The agent parses the HTML itself, or you build a template in the visual editor - load the page, click the fields, publish - and the app calls it by name. A template can be fixed in the editor when the site changes, without touching the app.
1 credit per successful fetch; failures are free. Each recipe has an estimate - a daily rank tracker for 100 keywords is about 3,000 credits a month. New accounts get 500 free credits, and plans make credits cheaper at volume.
Fetch public pages, respect each site's terms, cache what you fetched and refresh no more often than the data changes. Use is governed by our Acceptable Use Policy.
Build your first scraper today.
500 free credits, no credit card. Connect your AI in a minute.