---
title: "Using Haystack Docs in Your Coding Agent"
id: docs-mcp-server
slug: "/docs-mcp-server"
description: "Connect your coding agent to the Haystack documentation through the public MCP server. Includes setup instructions for Claude Code, Cursor, and GitHub Copilot."
---

import Tabs from '@theme/Tabs';
import TabItem from '@theme/TabItem';

# Using Haystack Docs in Your Coding Agent

Haystack publishes a public [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server that lets coding agents search the official Haystack documentation. Pointing your agent at it means it answers questions from up-to-date docs instead of relying on training data, which can lag behind the framework.

The server exposes a single tool, `search_haystack_docs`, that returns relevant documentation sections with source URLs. No API key or sign-up is needed.

## Server URL

```
https://docs.haystack.deepset.ai/api/mcp
```

The server speaks **HTTP** transport. Most agents auto-detect this. If yours asks you to choose, pick `http`.

## Setup

<Tabs>
<TabItem value="claude-code" label="Claude Code" default>

Add the server with the `claude mcp` CLI:

```shell
claude mcp add --transport http haystack-docs https://docs.haystack.deepset.ai/api/mcp
```

Restart your Claude Code session. Verify it's connected by running `/mcp` — you should see `haystack-docs` listed with the `search_haystack_docs` tool.

For more options (project vs. user scope, SSE transport, headers), see the [Claude Code MCP docs](https://code.claude.com/docs/en/mcp).

</TabItem>
<TabItem value="cursor" label="Cursor">

Open Cursor settings → **Tools & MCPs** → **Add new MCP server**, or edit `~/.cursor/mcp.json` (global) or `.cursor/mcp.json` (per-project) directly:

```json
{
  "mcpServers": {
    "haystack-docs": {
      "url": "https://docs.haystack.deepset.ai/api/mcp"
    }
  }
}
```

Save the file and reload Cursor. The tool appears in the **Available Tools** list inside the chat panel.

See the [Cursor MCP docs](https://cursor.com/docs/context/mcp) for the full configuration reference.

</TabItem>
<TabItem value="copilot" label="GitHub Copilot">

In VS Code, create or edit `.vscode/mcp.json` in your workspace:

```json
{
  "servers": {
    "haystack-docs": {
      "type": "http",
      "url": "https://docs.haystack.deepset.ai/api/mcp"
    }
  }
}
```

Open the Copilot Chat panel, switch to **Agent** mode, then click the tools icon and enable `haystack-docs`. You can also register the server globally from the command palette via **MCP: Add Server**.

See [Add and manage MCP servers in VS Code](https://code.visualstudio.com/docs/copilot/chat/mcp-servers) for the full configuration reference.

</TabItem>
</Tabs>

## Verifying it works

Ask your agent a question that requires current Haystack knowledge, for example:

> What are the required methods on a Haystack custom component?

The agent should call `search_haystack_docs` and cite source URLs under `docs.haystack.deepset.ai` in its answer. If it answers without calling the tool, prompt it explicitly: *"Use the haystack-docs MCP server to answer."*