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

# How to Use nTop's MCP Server

### **Objective**

This article explains how to connect AI applications to your technical content using nTop's Model Context Protocol (MCP) server.

### **Applies to**

Developers and technical professionals using AI agents and the nTop knowledge base.

## **Overview**

The nTop MCP server provides tools that allow AI applications to search and retrieve information from the nTop knowledge base, learning center, and other guides. By using this server, you can integrate nTop's technical documentation directly into your AI-driven workflows.

Hosted MCP Server: [https://docs.ntop.com/mcp](https://docs.ntop.com/mcp)

### **Available Tools**

The server exposes two primary tools to connected AI clients:

* **search\_n\_top:** Use this tool to find relevant information, examples, and references. It returns contextual titles and direct links to documentation pages.
* **query\_docs\_filesystem\_n\_top:** Use this tool to run shell-like queries (such as ls, cat, or rg) against a virtualized filesystem containing nTop documentation.

### **Procedure**

Please follow these steps to access and utilize the MCP server:

**1. Access the Hosted MCP Server**

1. Navigate to the **Agents** or **MCP** section in your application settings.
2. Locate the **Hosted MCP server** URL.
3. Use this URL to connect your AI application (e.g., Claude Desktop, Custom GPT, or a local agent) to nTop content.

**2. Search for Information**

When you need to answer broad questions or locate specific features, we recommend starting with the search tool:

1. Use the **search\_n\_top** tool for conceptual queries (e.g., "How do I run an aircraft flow analysis?").
2. The tool returns a list of relevant paths, such as **/knowledge-base/fluids/how-to-run-an-aircraft-flow-analysis**.

**3. Retrieve Specific Content**

Once you have a documentation path, use the filesystem tool to read the full content:

1. Identify the file path returned from your search
2. Append .mdx to the path.
3. Use commands like head or cat to read the file.
   * Example: cat /knowledge-base/fluids/how-to-run-an-aircraft-flow-analysis.mdx
4. If you need to find specific technical requirements (like GPU compatibility) across multiple files, use the rg (ripgrep) command.

**4. Optimize Your Workflow**

To ensure efficient performance and stay within data limits:

* **Search first:** Always use **search\_n\_top** before querying the filesystem to find the exact path.
* **Batch reads:** We recommend reading multiple sections in a single call using **head** or **cat** when possible.
* **Use targeted queries:** Prefer **rg -C** (context search) or **head** over broad **cat** commands on large files to keep output under the 30KB limit.

And that's it! You've successfully connected and queried the nTop MCP server.

Are you still having issues? Contact the nTop support team, and we'll be happy to help!

###