MCP LLMS-TXT Documentation Server

Overview

llms.txt is a standard index of website contents to help LLMs. As an example, LangGraph's llms.txt provides a curated list of LangGraph doc URLs with a short description of each one. An LLM can use this file to decide which pages to read when accomplishing tasks, and pairs well with IDEs like Cursor and Windsurf or applications like Claude Code/Desktop.

However, these applications use different built-in tools to read and process files like llms.txt; sometimes IDEs will reflect on the llms.txt file and use it for formulate web search queries rather than retrieving the specific URLs listed! More broadly, there can be poor visibility into what applications are doing with their built-in retrieval / search tools.

MCP offers a way for developers to define tools that give us full control over how documentation is retrieved and displayed to LLMs in these applications. Here, we create a simple MCP server that defines a few basical external tools that these applications can use: 1) to tool to load llms.txt and 2) fetch specific URLs within llms.txt. When these tools are used, the user can customize retrieval and audit the tool calls / the context returned to better understand what is happening under the hood.

Quickstart

Install uv:

curl -LsSf https://astral.sh/uv/install.sh | sh

Please see official uv docs for other ways to install uv.

Select an llms.txt file to use. For example, here's the LangGraph llms.txt

https://langchain-ai.github.io/langgraph/llms.txt  

Run the MCP server locally with whatever llms.txt file you want to use:

uvx --from mcpdoc mcpdoc \
    --urls LangGraph:https://langchain-ai.github.io/langgraph/llms.txt \
    --transport sse \
    --port 8081 \
    --host localhost

Run MCP inspector and connect to the running server via SSE at http://localhost:8081/sse:

npx @modelcontextprotocol/inspector

Here, you can test the tool calls.

Finally, add the server to any MCP host applications of interest.

Below, we walk through each one, but here are the the config files that are updated for each:

*Cursor*
`~/.cursor/mcp.json` 

*Windsurf*
`~/.codeium/windsurf/mcp_config.json`
 
*Claude Desktop*
`~/Library/Application\ Support/Claude/claude_desktop_config.json`
 
*Claude Code*
`~/.claude.json`

These will be updated with our server specification, as shown below.

NOTE: It appears that stdio transport required for Windsurf and Cursor.

{
  "mcpServers": {
    "langgraph-docs-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "mcpdoc",
        "mcpdoc",
        "--urls",
        "LangGraph:https://langchain-ai.github.io/langgraph/llms.txt",
        "--transport",
        "stdio",
        "--port",
        "8081",
        "--host",
        "localhost"
      ]
    }
  }
}

Usage

Cursor

Setup:

• Ensure ~/.cursor/mcp.json is updated to include the server.
• Settings -> MCP to confirm that the server is connected.
• Control-L to open chat.
• Ensure agent is selected.

Then, try an example prompt:

use the langgraph-docs-mcp server to answer any LangGraph questions -- 
+ call get_docs tool to get the available llms.txt file
+ call fetch_docs tool to read it
+ reflect on the urls in llms.txt 
+ reflect on the input question 
+ call fetch_docs on any urls relevant to the question
+ use this to answer the question

what are types of memory in LangGraph?

• It will ask to approve tool calls.

Windsurf

Setup:

• Ensure ~/.codeium/windsurf/mcp_config.json is updated to include the server.
• Control-L to open Cascade.
• Available MCP servers will be listed.

Then, try the example prompt:

• It will perform your tool calls.

Claude Desktop

Setup:

• Open Settings -> Developer to update the config.
• Restart Claude.

• You will see your tools.

Then, try the example prompt:

• It will ask to approve tool calls.

Claude Code

Setup:

• Shortcut to add the MCP server to your project:

claude mcp add-json langgraph-docs '{"type":"stdio","command":"uvx" ,"args":["--from", "mcpdoc", "mcpdoc", "--urls", "langgraph:https://langchain-ai.github.io/langgraph/llms.txt"]}' -s project

• Test

$ Claude
$ /mcp 

Then, try the example prompt:

• It will ask to approve tool calls.

Command-line Interface

The mcpdoc command provides a simple CLI for launching the documentation server. You can specify documentation sources in three ways, and these can be combined:

1. Using a YAML config file:

mcpdoc --yaml sample_config.yaml

This will load the LangGraph Python documentation from the sample_config.yaml file.

2. Using a JSON config file:

mcpdoc --json sample_config.json

This will load the LangGraph Python documentation from the sample_config.json file.

3. Directly specifying llms.txt URLs with optional names:

mcpdoc --urls https://langchain-ai.github.io/langgraph/llms.txt LangGraph:https://langchain-ai.github.io/langgraph/llms.txt

URLs can be specified either as plain URLs or with optional names using the format name:url.

You can also combine these methods to merge documentation sources:

mcpdoc --yaml sample_config.yaml --json sample_config.json --urls https://langchain-ai.github.io/langgraph/llms.txt

Additional Options

• --follow-redirects: Follow HTTP redirects (defaults to False)
• --timeout SECONDS: HTTP request timeout in seconds (defaults to 10.0)

Example with additional options:

mcpdoc --yaml sample_config.yaml --follow-redirects --timeout 15

This will load the LangGraph Python documentation with a 15-second timeout and follow any HTTP redirects if necessary.

Configuration Format

Both YAML and JSON configuration files should contain a list of documentation sources. Each source must include an llms_txt URL and can optionally include a name:

YAML Configuration Example (sample_config.yaml)

# Sample configuration for mcp-mcpdoc server
# Each entry must have a llms_txt URL and optionally a name
- name: LangGraph Python
  llms_txt: https://langchain-ai.github.io/langgraph/llms.txt

JSON Configuration Example (sample_config.json)

[
  {
    "name": "LangGraph Python",
    "llms_txt": "https://langchain-ai.github.io/langgraph/llms.txt"
  }
]

Programmatic Usage

from mcpdoc.main import create_server

# Create a server with documentation sources
server = create_server(
    [
        {
            "name": "LangGraph Python",
            "llms_txt": "https://langchain-ai.github.io/langgraph/llms.txt",
        },
        # You can add multiple documentation sources
        # {
        #     "name": "Another Documentation",
        #     "llms_txt": "https://example.com/llms.txt",
        # },
    ],
    follow_redirects=True,
    timeout=15.0,
)

# Run the server
server.run(transport="stdio")