Yellow.ai Docs MCP Server

Connect your AI assistant — Claude, ChatGPT, Cursor, or any other MCP-capable tool — directly to Yellow.ai's product documentation. Once connected, your assistant can search and read docs.yellow.ai on demand and answer your Yellow.ai questions with accurate, cited information instead of guessing.

It's a public Model Context Protocol (MCP) server.

note

No login, no API key, no setup on your Yellow.ai account. The server reads only public documentation — there is nothing to authorize, and nothing it can see or change on your account.

Why use it

• Grounded, cited answers. Your assistant quotes the actual docs and links the exact page, instead of relying on stale or approximate training data.
• Always current. The index re-crawls docs.yellow.ai continuously, so new and updated documentation shows up automatically.
• Build faster. Keep it loaded in your IDE assistant (Cursor, Claude Code) while you build on the platform — ask "how do I configure X?" and get the exact steps without leaving your editor.

What it can do

The server exposes two standard tools that your AI client uses automatically:

Tool	What it does
search	Finds the most relevant documentation pages for your query.
fetch	Returns the full content of a specific page for the assistant to read and cite.

You never call these by hand — your AI client discovers them and uses them whenever you ask a Yellow.ai question.

Connect your client

The server is available at:

https://nexus.yellow.ai/docs-mcp/mcp

note

It's hosted under nexus.yellow.ai rather than docs.yellow.ai for now, for infrastructure reasons — the URL will be announced here if that ever changes. Bookmark this page rather than the raw URL.

Add it to your MCP client using the instructions for that client below. All of them point at the same URL — the server speaks Streamable HTTP, the standard remote-MCP transport, so any MCP client works.

Claude Code (CLI)

claude mcp add --transport http yellow-docs https://nexus.yellow.ai/docs-mcp/mcp

Or add it to your project's .mcp.json:

{
  "mcpServers": {
    "yellow-docs": {
      "type": "http",
      "url": "https://nexus.yellow.ai/docs-mcp/mcp"
    }
  }
}

Claude.ai and Claude Desktop

Open Connectors in Claude's settings — Customize → Connectors on claude.ai, or Settings → Connectors in Claude Desktop — choose Add custom connector, and paste https://nexus.yellow.ai/docs-mcp/mcp.

note

On Team and Enterprise plans, an administrator may need to enable custom connectors for your workspace.

Cursor

Add the server to ~/.cursor/mcp.json (global) or .cursor/mcp.json (per project):

{
  "mcpServers": {
    "yellow-docs": {
      "url": "https://nexus.yellow.ai/docs-mcp/mcp"
    }
  }
}

VS Code (Copilot agent mode)

Add it to .vscode/mcp.json:

{
  "servers": {
    "yellow-docs": {
      "type": "http",
      "url": "https://nexus.yellow.ai/docs-mcp/mcp"
    }
  }
}

ChatGPT (deep research and Apps)

In a workspace with custom apps enabled, add a custom MCP connector pointing at https://nexus.yellow.ai/docs-mcp/mcp — under Settings → Apps (formerly "Connectors"). The server's search/fetch shape is exactly what ChatGPT's deep research and company-knowledge expect, so no extra configuration is needed.

Any other MCP client

Any client that speaks Streamable HTTP works — point it at https://nexus.yellow.ai/docs-mcp/mcp. No credentials are required.

Try it

After connecting, restart your client and ask a normal question, for example:

• "Using the Yellow.ai docs, how do I create my first AI agent?"
• "What channels can a Yellow.ai bot be deployed on, and where do I configure WhatsApp?"
• "How do I generate a bot API key in Yellow.ai?"
• "Walk me through setting up a voice agent, and link the docs you used."

Your assistant will search the documentation, fetch the most relevant page, and answer with a link back to docs.yellow.ai so you can verify the source.

Get the most out of it

• Ask for citations — add "…and link the exact doc page" so you can confirm the answer against the source.
• Name the product area ("in Nexus…", "for the chat widget…", "for voice…") to sharpen the search.
• Chain your questions — follow-ups let the assistant fetch deeper or related pages as the conversation develops.
• Pair it with building — while wiring up an integration, flow, or agent, ask for the configuration steps inline in your IDE assistant instead of switching tabs.

Good to know

• Read-only and public. The server indexes only public documentation. It cannot see or change your bot, your data, or your account.
• Freshness. Brand-new documentation may take a short while to appear as the index refreshes; if a page is very new, try again shortly.
• Keyword search. If a search misses, rephrase using the term the docs use (for example, "handover" rather than "handoff").

Related

• New to the platform? See how to use this documentation for the recommended reading order.
• Looking to let your bot's AI Agent call third-party MCP servers (HubSpot, Postgres, and others) instead? That's a different feature — see the MCP Data Connector.