Open-source MCP server

Give your AI a real browser.

Your AI assistant keeps hitting “you're blocked” walls. @unblockingapi/mcp connects Claude, Cursor, VS Code or Zed to UnblockingAPI's proxies and rendering engine - so they read pages the way you see them.

View source

500 free credits  ·  No credit card  ·  MIT-licensed

amazon.comSample data

Select one item. Similar items are detected automatically.

Website

Sony WH-1000XM5

4.6

$299.99

Bose QuietComfort

4.5

$249.00

Bang & Olufsen Beoplay

4

$1,714.99

Extracted JSON

{ "products": [
  { "name": "Sony WH-1000XM5", "price": 299.99, "currency": "USD", "rating": 4.6 },
  { "name": "Bose QuietComfort", "price": 249, "currency": "USD", "rating": 4.5 },
  { "name": "Bang & Olufsen Beoplay", "price": 1714.99, "currency": "USD", "rating": 4 }
] }

Tap one item, get every item - as an array.

Try it live

Three steps

Set up in about a minute

No installer, no browser extension - a key and a config snippet.

Step 1

Get an API key

500 free credits, no credit card. Takes a minute to sign up.

Step 2

Add the MCP server

One command or a short config snippet, depending on your tool.

Step 3

Ask it to fetch a page

Your assistant reaches for unblock_fetch on its own, and for a template when one fits.

Pick your assistant

Same server, same API key. Connect as many as you use.

Claude

A Claude Code plugin that makes unblock_fetch Claude's default web fetch; Claude Desktop gets the same tools over MCP.

Plugin, CLI command or config file

Cursor

Cursor's agent reads live docs and pages instead of guessing from memory.

Drop into mcp.json

VS Code

GitHub Copilot's agent mode gets pages that normally refuse automated tools.

code --add-mcp, or mcp.json

Zed

Zed's agent panel gains a fetch tool that runs through real proxies.

Add to settings.json
For developers

There's a raw HTTP API too

The MCP server is a thin wrapper around UnblockingAPI's own endpoint. Call it directly from any script or backend and get clean HTML back.

Same account, same credits, same key - full API reference.

curl -H "X-Api-Key: YOUR_API_KEY" \
  "https://api.unblockingapi.com/unblock?url=https://example.com"

Add &render=true to run a real headless browser and get the JavaScript-rendered page back.

No code

Want structured data instead of raw HTML?

unblock_fetch hands back the full rendered page. For clean fields instead - a price, a title, a whole list - build a template visually in the editor: load the page, tap what you want, done. No CSS selectors to write by hand.

Save it and it's a reusable API endpoint - same account, same key, same as the built-in templates. Build the structure once, then call it from your app - or your AI agent - forever.

Try the editor

No account needed to try it  ·  No credit card  ·  More about the editor

amazon.comSample data

Select one item. Similar items are detected automatically.

Website

Sony WH-1000XM5

4.6

$299.99

Bose QuietComfort

4.5

$249.00

Bang & Olufsen Beoplay

4

$1,714.99

Extracted JSON

{ "products": [
  { "name": "Sony WH-1000XM5", "price": 299.99, "currency": "USD", "rating": 4.6 },
  { "name": "Bose QuietComfort", "price": 249, "currency": "USD", "rating": 4.5 },
  { "name": "Bang & Olufsen Beoplay", "price": 1714.99, "currency": "USD", "rating": 4 }
] }

Tap one item, get every item - as an array.

What actually gets you past the block

Every fetch runs through the same infrastructure whether it comes from the MCP server, the raw API, or the Swagger UI.

Rotating residential proxies

Requests go out through real residential IPs and an anti-detection engine, so sites that block scripted tools generally load the way they would in a browser.

Real JavaScript rendering

Ask for a rendered fetch and a real headless browser runs the page before returning the DOM - SPAs and dashboards included.

Open-source client

The MCP server itself is MIT-licensed and public on GitHub and npm - inspect exactly what it sends before you trust it with a key.

Prove it works in one prompt

Once you've connected a tool, paste this into its chat. Google normally turns robots away.

Use unblockingapi to fetch a product page from amazon.com as structured data - a title, a price and a rating.

Works everywhere

Any client that speaks MCP over stdio, on any OS

500 free credits

No credit card required to sign up and try it

Open source

MIT-licensed - read exactly what it sends before you trust it

Questions

Yes - every request goes through UnblockingAPI's proxy and rendering infrastructure, which needs a key. New accounts start on Pay as you Go with 500 free credits and no credit card required.

@unblockingapi/mcp exposes UnblockingAPI to your AI assistant as tools it can call directly - fetching a URL, running a Google search, or pulling structured data from supported page types.

Claude Code, Claude Desktop, Cursor, VS Code (Copilot agent mode) and Zed all run it the same way - as a local process launched with npx. Anything else that speaks MCP over stdio works the same way.

Yes - the MCP server is a thin wrapper around it. The API docs cover the full endpoint if you'd rather call it directly from a script.

Yes, MIT-licensed. Source on GitHub and the package on npm.

Requests return a clear 402 error rather than failing silently, so your assistant knows exactly why a fetch didn't work. Top up or upgrade any time.

Building your own agent or RAG pipeline? See the AI agents & LLM apps use case.

Give your AI a real browser.

500 free credits, no credit card. Connect in under a minute.