DEV Community

Cover image for Building a free backlink checker as a solo dev — with MCP support for Claude
Mouhab Mzibra
Mouhab Mzibra

Posted on

Building a free backlink checker as a solo dev — with MCP support for Claude

## The problem

I've been building seo-backlinks.net — a free backlink checker — solo from Morocco for the past year. Last month I asked myself a stupid question: can Claude actually
check backlinks for me while I'm coding?

Turns out yes. And the answer is MCP (Model Context Protocol).

## What I built

An MCP server that lets Claude Desktop or Claude Code query 200M+ backlinks from Common Crawl data in real time:

▎ Claude, check the backlink profile of stripe.com
▎ Claude, find expired domains with DR > 40 in the SEO niche
▎ Claude, who links to my competitor's homepage?
## The stack (nothing fancy)

  • Data: Common Crawl monthly dumps → parsed link graph
  • Storage: Postgres 16 on a €6/mo AlexHost VPS (yes, really)
  • API: Bun HTTP server behind Caddy
  • MCP layer: stdio transport, 5 tools exposed
  • Auth: API key from web dashboard

Total infra cost: ~€8/month for the whole thing serving real users.

## The MCP tools

I ship 5 tools with the server:

| Tool | What it does |
|---|---|
| check_domain | Full backlink profile for any domain |
| get_referring_domains | List domains linking to a target |
| get_top_hosts | Top backlink sources ranked by authority |
| get_expired_domains | Find drops with existing backlink equity |
| get_domain_info | Domain rating, spam score, category |

## Install it in Claude Code


bash
  claude mcp add seo-backlinks -- npx -y seo-backlinks-mcp

  You get a free tier (10 lookups/day, no card). Grab an API key at https://seo-backlinks.net.

  What I learned building this  

  1. MCP is stupidly easy. The whole server is <300 LOC. If you have an API, wrapping it in MCP is a weekend project.

  2. Users will find your API before you find users. My backend was getting real hits from a Morocco IP for weeks before I even had a landing page — turns out you (my future self testing) count as a user.

  3. Common Crawl is a superpower. 100TB+ of web link data, updated monthly, free. Everyone should have a hobby project using it.

  What's next

  - Chrome extension (in progress)
  - CLI (npx check-backlinks stripe.com)
  - Vercel AI SDK integration for JS devs

  If you build with Claude and care about SEO, try it. If you're an SEO curious about AI tooling, also try it. Feedback + issues welcome on GitHub (https://github.com/mouhabmzibra-rgbs/mcp-seo-backlinks).

  Solo devs unite 🇲🇦  
Enter fullscreen mode Exit fullscreen mode

Top comments (0)