Start free. Scale as your blog grows.Get started

Build with AI

Docs MCP server

Give your AI coding assistant the ContioReach docs. Connect the Docs MCP server to Cursor, Claude Code, VS Code or Claude, and your assistant looks up the real API, webhook rules and framework guides before it writes a line of code.

It is free, needs no account and no API key, and only ever reads the public documentation on this site.

Server details

SettingValueNotes
Namecontioreach-docsAny name works; this one is used in the examples below.
URLhttps://mcp.contioreach.com/docsThe only thing most clients ask for.
TransportStreamable HTTPSometimes labelled "HTTP" or "remote".
AuthenticationNonePublic, read-only docs. Choose "No sign in" if asked.

Install

Pick your tool. Any client that supports remote MCP servers works.

Terminal
claude mcp add --transport http contioreach-docs https://mcp.contioreach.com/docs

Start a new session and run /mcp: contioreach-docs should show as connected with three tools.

See it work

Once connected, you don't call the tools yourself. Ask for what you want; the assistant decides when it needs the docs.

Cursor · my-nuxt-blog

Add ContioReach webhooks to my Nuxt app so posts update without a redeploy.

contioreach-docssearch_docs("webhooks", framework: "nuxt")

3 results: Nuxt guide › Publish instantly with a webhook · Webhooks › Configure your webhook · Webhooks › When webhooks are sent

contioreach-docsget_doc_page("webhooks")

Webhooks — the full page as Markdown

Added server/api/revalidate/all.post.js. It checks the secret in the X-API-Key header against REVALIDATION_SECRET and clears the cached posts, so a published post shows up on the next visit.

Why connect it

Without the Docs MCP

ContioReach is newer than most models' training data. Expect invented endpoints, the API key in a public NEXT_PUBLIC_ variable, or a webhook route that never checks its secret.

With the Docs MCP

The assistant reads the current docs first, so the code uses the real response shapes, keeps the key on the server and follows the webhook rules. Docs changes reach it as soon as they are published.

Tools

The server exposes three read-only tools.

ToolArgumentsWhat it does
search_docsquery, framework?, limit?Finds the most relevant sections across every docs page and returns them with links. framework (nextjs, nuxt, astro, sveltekit, remix, react, vue, gatsby) puts that guide first and leaves the other frameworks' guides out.
get_doc_pagepageReturns one full page as Markdown, code samples and tables included. Accepts a path such as /docs/api/posts, a URL, or a short name such as webhooks or nuxt.
list_doc_pages—Lists every docs page with its path and section headings.

Keep your client's tool-approval setting on, as you would for any MCP server, so you can see each call before it runs. These tools only read public docs, so approving them is low-risk.

Try asking

  • Build a blog index page in Next.js with pagination from ContioReach.
  • Add category archive pages to my SvelteKit blog.
  • Why does my post page show no reading time?
  • Handle the ContioReach 429 usage-limit error gracefully.
  • Set up the publish webhook in my Astro site and test it with curl.

Make your assistant use it every time

Assistants call the server on their own when a question is clearly about ContioReach. To make sure they check the docs on every related change, add one line to your project's instructions file — CLAUDE.md, AGENTS.md, or a Cursor rule in .cursor/rules:

AGENTS.md
When working with ContioReach (the CMS API, webhooks or the blog starters), look up the current docs with the contioreach-docs MCP server before writing code.

Questions

Do I need a ContioReach account?

No. The server only reads our public documentation, so there is nothing to sign in to.

Can it see my posts?

No. It never touches a workspace or your API key. Your own content is read by your site through the Content API, as described in the quickstart.

Is it free?

Yes, with or without a ContioReach plan.

Does the AI company store the docs?

No. Your assistant fetches the sections it needs, when it needs them, from our server. Nothing is uploaded or used for training through MCP.

It doesn't connect

  • Check the URL is exactly https://mcp.contioreach.com/docs and the transport is HTTP, not SSE or stdio.
  • Restart the client after adding the server; most only read their MCP settings at startup.
  • On a company laptop, ask IT to allow mcp.contioreach.com — managed networks sometimes block MCP servers.

Ready to connect real content?

Was this page helpful?