For AI agents & developers

The Bikefinest MCP

Bikefinest runs a public Model Context Protocol server. Point any MCP-capable AI agent at it and ask cycling questions in plain language — it searches our entire knowledge base and returns the most relevant reviews, RideScore™rankings, head-to-head comparisons, how-to guides, and community Q&A, each with source links back to the page.

The endpoint

Streamable HTTP. Public, no authentication. Rate limited to 120 requests per minute.

https://ae7d2747-0916-4547-b662-75465579fd32.search.ai.cloudflare.com/mcp

What it can answer

The server indexes the whole site — 86 products across 11 categories, 294 head-to-head comparisons, 54 ranked best-of lists, plus how-to guides, news and community intel — and stays in sync automatically as the site updates. Every answer is grounded in the published RideScore™ methodology.

Connect your client

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "bikefinest": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://ae7d2747-0916-4547-b662-75465579fd32.search.ai.cloudflare.com/mcp"]
    }
  }
}

Cursor

Add to ~/.cursor/mcp.json (Cursor supports remote URLs directly):

{
  "mcpServers": {
    "bikefinest": {
      "url": "https://ae7d2747-0916-4547-b662-75465579fd32.search.ai.cloudflare.com/mcp"
    }
  }
}

Any remote-MCP client / Cloudflare AI Playground

Paste the endpoint URL into any client that supports remote MCP over streamable HTTP (for a no-install test, use the Cloudflare AI Playground and add the URL under MCP Servers).

The tool

The server exposes one tool, search, which takes a natural-language query and returns ranked, source-linked content. Example prompts once connected:

“What’s the best road bike for a beginner under $1,500?”
“Compare the Trek Domane and the Specialized Roubaix.”
“Which tyre should I run for wet gravel, and why?”
“What do owners actually say about the DT Swiss ER 1600 wheelset?”

Good to know

Free and open. No key, no sign-up. The machine-readable server card lives at /.well-known/mcp/server-card.json.
It retrieves; you decide. The server surfaces ranked options and evidence — it never buys anything or makes the call for you. Purchase decisions stay yours.
Always current. The index re-syncs from our sitemap, so answers reflect the live site.
Open source. Reference implementation, companion skill, and connect docs: github.com/automationsquadhq/bikefinest-mcp.