> ## Documentation Index
> Fetch the complete documentation index at: https://docs.lynxhub.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Model Context Protocol

> Connect AI tools like Claude, Cursor, and VS Code to LynxHub's hosted search MCP server.

LynxHub hosts an open [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) search server at `https://docs.lynxhub.app/mcp`.

Connecting your AI application to the LynxHub MCP server allows your model to interactively search, browse, and retrieve up-to-date LynxHub documentation while generating responses.

## Hosted MCP URL

You can connect your AI client to LynxHub's public search MCP server using:

* **MCP Endpoint:** [`https://docs.lynxhub.app/mcp`](https://docs.lynxhub.app/mcp)
* **Discovery Endpoint:** `https://docs.lynxhub.app/.well-known/mcp`

## Available MCP tools and resources

When connected, your AI agent gains access to the following capabilities:

### Tools

* **`search`:** Searches across all LynxHub documentation pages to find relevant snippets and topic URLs matching a query.
* **`query docs filesystem`:** Reads virtual documentation directories and fetches full Markdown contents of specific LynxHub pages in a single batch call.
* **`submit feedback`:** Allows AI agents to report outdated or missing documentation back to the LynxHub team.

### Resources

* **`skill.md`:** Exposes [LynxHub skills](/ai-agents/skill-md) as MCP resources so your agent can inspect actionable workflows without additional CLI setup.

## Connect AI tools to the LynxHub MCP server

Follow the setup instructions below for your preferred AI tool or editor:

<Tabs>
  <Tab title="Claude">
    To connect LynxHub documentation to Claude:

    <Steps>
      <Step title="Open connector settings">
        Navigate to the **Connectors** section in your [Claude Settings](https://claude.ai/settings/connectors).
      </Step>

      <Step title="Add custom connector">
        Click **Add custom connector** and fill in the details:

        * **Name:** `LynxHub`
        * **URL:** `https://docs.lynxhub.app/mcp`
      </Step>

      <Step title="Select in chat">
        In your Claude chat window, click the attachments icon (+), select **LynxHub**, and ask your question.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Claude Code">
    To add the LynxHub MCP server to Claude Code, run the following terminal command:

    ```bash theme={null}
    claude mcp add --transport http LynxHub https://docs.lynxhub.app/mcp
    ```

    Verify the connection by running:

    ```bash theme={null}
    claude mcp list
    ```
  </Tab>

  <Tab title="Cursor">
    To add LynxHub search to Cursor:

    <Steps>
      <Step title="Open MCP settings">
        Press `Ctrl+Shift+P` (or `Cmd+Shift+P` on macOS), search for `Open MCP settings`, and select **Add custom MCP**.
      </Step>

      <Step title="Configure mcp.json">
        Add the `LynxHub` server configuration to your `mcp.json` file:

        ```json theme={null}
        {
          "mcpServers": {
            "LynxHub": {
              "url": "https://docs.lynxhub.app/mcp"
            }
          }
        }
        ```
      </Step>

      <Step title="Test in Cursor Chat">
        Ask Cursor Chat: *"How do I configure terminal hotkeys in LynxHub?"* Cursor will automatically call the LynxHub search MCP tool.
      </Step>
    </Steps>
  </Tab>

  <Tab title="VS Code">
    To connect LynxHub MCP in VS Code:

    1. Create or open `.vscode/mcp.json` in your workspace.
    2. Add the LynxHub server entry:

    ```json theme={null}
    {
      "servers": {
        "LynxHub": {
          "type": "http",
          "url": "https://docs.lynxhub.app/mcp"
        }
      }
    }
    ```
  </Tab>

  <Tab title="Codex CLI">
    To connect LynxHub MCP to Codex CLI:

    Open `~/.codex/config.toml` and add the following configuration block:

    ```toml theme={null}
    [mcp_servers.lynxhub]
    url = "https://docs.lynxhub.app/mcp"
    ```
  </Tab>
</Tabs>

## Which AI integration to use and when

LynxHub provides three complementary AI integration standards. Refer to the matrix below to choose the best option for your workflow:

| Feature             | `llms.txt`                            | `skill.md`                          | MCP Server                               |
| :------------------ | :------------------------------------ | :---------------------------------- | :--------------------------------------- |
| **Primary Purpose** | Documentation index for LLMs          | Actionable capabilities & workflows | Real-time interactive doc search         |
| **Access URL**      | `https://docs.lynxhub.app/llms.txt`   | `https://docs.lynxhub.app/skill.md` | `https://docs.lynxhub.app/mcp`           |
| **Best For**        | Static indexing & RAG pipelines       | Plugin development & task execution | Coding in IDEs (Cursor, VS Code, Claude) |
| **Context Usage**   | Low (index) to High (`llms-full.txt`) | Low (loaded on demand)              | Zero context overhead until searched     |
| **Interactivity**   | Passive read                          | Guided step-by-step                 | Active tool calling & batch queries      |

<Tip>
  **Pro-tip:** For the best coding experience in Cursor or VS Code, connect the **MCP Server** for real-time doc lookups, and load **`skill.md`** when building custom
  LynxHub extensions.
</Tip>