LlamaIndex Documentation MCP server

The LlamaIndex MCP Server allows you to create powerful, local Model Context Protocol servers that connect AI models like Claude to your private data sources through retrieval-augmented generation (RAG). This enables AI assistants to access and reference your specific information when answering your questions.

Installation and Setup

Prerequisites

• LlamaCloud account
• Claude Desktop (for client usage)
• OpenAI API key (or alternative LLM)
• Python with Poetry (recommended)

Setting Up Your LlamaCloud Index

1. Create an account on LlamaCloud
2. Create a new index through the UI and add your data source:
   • You can use Google Drive, direct document uploads, or other sources
   • For testing, you might want to upload a small set of documents directly
3. Obtain your API key from the LlamaCloud UI dashboard

Installing the MCP Server

1. Clone the repository
2. Create a .env file in the root directory with the following variables:

   LLAMA_CLOUD_API_KEY=your_llama_cloud_api_key
   OPENAI_API_KEY=your_openai_api_key
   

Using the MCP Server with Claude Desktop

Starting the MCP Server

The server uses a simple Python script with the following structure:

from mcp_server import FastMCP
import os

mcp = FastMCP('llama-index-server')

@mcp.tool()
def llama_index_documentation(query: str) -> str:
    """Search the llama-index documentation for the given query."""
    # Initialize your LlamaCloud index
    index = LlamaCloudIndex(
        name="your-index-name",
        project_name="Your Project",
        organization_id="your-org-id",
        api_key=os.getenv("LLAMA_CLOUD_API_KEY"),
    )
    
    # Execute the query
    response = index.as_query_engine().query(query + " Be verbose and include code examples.")
    
    return str(response)

if __name__ == "__main__":
    mcp.run(transport="stdio")

Configuring Claude Desktop

1. Open Claude Desktop
2. Go to Claude → Settings → Developer → Edit Config
3. Configure your MCP server in the JSON config:

{
    "mcpServers": {
        "llama_index_docs_server": {
            "command": "poetry",
            "args": [
                "--directory",
                "/path/to/your/llamacloud-mcp",
                "run",
                "python",
                "/path/to/your/llamacloud-mcp/mcp-server.py"
            ]
        }
    }
}

4. Restart Claude Desktop
5. You should now see your tool listed under the query box in Claude Desktop

Using LlamaIndex as an MCP Client

Setting Up an HTTP MCP Server

For HTTP-based MCP access, you'll need to use a slightly modified server:

from mcp_server import FastMCP
import asyncio

mcp = FastMCP('llama-index-server', port=8000)

# Add your tools with @mcp.tool() decorator here

if __name__ == "__main__":
    asyncio.run(mcp.run_sse_async())

Creating a Client Application

from llama_index.multi_modal_llms.openai import OpenAI
from llama_index.agent import FunctionAgent
from llama_index.tools.mcp import BasicMCPClient, McpToolSpec
import asyncio

# Connect to your MCP server
mcp_client = BasicMCPClient("http://localhost:8000/sse")
mcp_tool_spec = McpToolSpec(
    client=mcp_client,
    # Optional: Filter tools by name
    # allowed_tools=["tool1", "tool2"],
)

# Get available tools
tools = mcp_tool_spec.to_tool_list()

# Create an agent with these tools
llm = OpenAI(model="gpt-4o-mini")
agent = FunctionAgent(
    tools=tools,
    llm=llm,
    system_prompt="You are an agent that knows how to build agents in LlamaIndex.",
)

# Run a query through the agent
async def run_agent():
    response = await agent.run("How do I instantiate an agent in LlamaIndex?")
    print(response)

if __name__ == "__main__":
    asyncio.run(run_agent())

Customizing Your MCP Tools

You can add multiple tools to your MCP server by creating additional decorated functions:

@mcp.tool()
def search_documentation(query: str) -> str:
    """Search the documentation for information."""
    # Implementation here
    
@mcp.tool()
def generate_code_example(description: str) -> str:
    """Generate a code example based on the description."""
    # Implementation here

Each tool will be available to the client (Claude Desktop or your custom application) as a separate function that can be invoked.

How to add this MCP server to Cursor

There are two ways to add an MCP server to Cursor. The most common way is to add the server globally in the ~/.cursor/mcp.json file so that it is available in all of your projects.

If you only need the server in a single project, you can add it to the project instead by creating or adding it to the .cursor/mcp.json file.

Adding an MCP server to Cursor globally

To add a global MCP server go to Cursor Settings > MCP and click "Add new global MCP server".

When you click that button the ~/.cursor/mcp.json file will be opened and you can add your server like this:

{
    "mcpServers": {
        "cursor-rules-mcp": {
            "command": "npx",
            "args": [
                "-y",
                "cursor-rules-mcp"
            ]
        }
    }
}

Adding an MCP server to a project

To add an MCP server to a project you can create a new .cursor/mcp.json file or add it to the existing one. This will look exactly the same as the global MCP server example above.

How to use the MCP server

Once the server is installed, you might need to head back to Settings > MCP and click the refresh button.

The Cursor agent will then be able to see the available tools the added MCP server has available and will call them when it needs to.

You can also explictly ask the agent to use the tool by mentioning the tool name and describing what the function does.