mcpdoc

To prevent unauthorized access to domains, mcpdoc implements strict domain access controls:

1. Remote llms.txt files: When you specify a remote URL (e.g., https://langchain-ai.github.io/langgraph/llms.txt), mcpdoc automatically adds only that specific domain (langchain-ai.github.io) to the allowed domains list.
2. Local llms.txt files: When using a local file, NO domains are automatically added. You MUST explicitly specify allowed domains using the --allowed-domains parameter.
3. Adding additional domains:
   • Use --allowed-domains domain1.com domain2.com to add specific domains.
   • Use --allowed-domains '*' to allow all domains (use with caution).

To use mcpdoc in Windsurf:

1. Open Cascade with CMD+L (Mac).
2. Click Configure MCP to open ~/.codeium/windsurf/mcp_config.json.
3. Add the server configuration (similar to the Cursor config).
4. Update Windsurf Rules/Global rules with instructions for the agent:

for ANY question about LangGraph, use the langgraph-docs-mcp server to help answer -- 
+ call list_doc_sources 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

The mcpdoc server is designed to be run using uv. You can install uv using the following command:

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

To add the mcpdoc server to Claude Code, run the following command in your terminal:

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

To verify the tools are available, launch Claude Code and run:

/mcp

Note: As of 3/21/25, you may need to append instructions to your prompt using <rules> tags to ensure the agent uses the tools correctly.

The mcpdoc server is configured using a list of documentation sources in either YAML or JSON format. Each source entry must include an llms_txt URL pointing to the documentation's llms.txt file. You may optionally include a name field to identify the source.

Required field:

• llms_txt: The URL of the documentation's llms.txt file.

Optional field:

• name: A human-readable name for the documentation source.

# YAML Example
- name: LangGraph Python
  llms_txt: https://langchain-ai.github.io/langgraph/llms.txt

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

You can test the mcpdoc server locally using uvx with the SSE transport. This allows you to run the server on a specific port and host for inspection.

uvx --from mcpdoc mcpdoc \
    --urls "LangGraph:https://langchain-ai.github.io/langgraph/llms.txt" "LangChain:https://python.langchain.com/llms.txt" \
    --transport sse \
    --port 8082 \
    --host localhost

After running, you can use the MCP inspector to connect and test tool calls:

npx @modelcontextprotocol/inspector

The following additional options are available for the mcpdoc CLI:

Example usage:

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

The mcpdoc command provides a CLI to launch the documentation server. You can specify documentation sources using YAML, JSON, or direct URLs.

1. Using a YAML config file:

mcpdoc --yaml sample_config.yaml

2. Using a JSON config file:

mcpdoc --json sample_config.json

3. Directly specifying URLs: URLs can be plain or formatted as name:url. You can use the --urls parameter multiple times to add multiple sources.

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

Combining methods: You can merge sources by using multiple flags:

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

The get_docs tool allows you to fetch LangGraph documentation from langchain-ai.github.io in markdown format.

Important Usage Pattern: You should always fetch the overview URL first. The overview response provides a list of available URLs that you can subsequently use to fetch specific documentation pages.

Arguments:

• url (str): The specific URL to fetch. It must either be the string "overview" or a valid URL starting with https://langchain-ai.github.io/.

# Example of how the tool is invoked via MCP
# 1. Get the overview to find available URLs
overview_content = await get_docs(url="overview")

# 2. Use a URL from the overview to get specific docs
# (Assuming 'https://langchain-ai.github.io/langgraph/concepts' was in the overview)
specific_docs = await get_docs(url="https://langchain-ai.github.io/langgraph/concepts")