LLM.txt MCP Server
A Model Context Protocol (MCP) server that extracts and serves context from llm.txt files, enabling AI models to understand file structure, dependencies, and code relationships in development environments. This server provides comprehensive access to the LLM.txt Directory, supporting file listing, content retrieval, and advanced multi-query search capabilities.
Features
- Directory listing with local caching (24-hour cache)
- OS-specific cache locations:
- Multi-query search with context
Installation
Recommended: Using MCP Get
The easiest way to install is using MCP Get, which will automatically configure the server in Claude Desktop:
Manual Configuration
Alternatively, you can manually configure the server in your Claude Desktop configuration by adding this to your
claude_desktop_config.json:Tools
1. list_llm_txt
Lists all available LLM.txt files from the directory. Results are cached locally for 24 hours.
Example response:
2. get_llm_txt
Fetches content from an LLM.txt file by ID (obtained from list_llm_txt).
Parameters:
id: The numeric ID of the LLM.txt file
Example response:
3. search_llm_txt
Search for multiple substrings within an LLM.txt file.
Parameters:
id: The numeric ID of the LLM.txt file
queries: Array of strings to search for (case-insensitive)
context_lines(optional): Number of lines to show before and after matches (default: 2)
Example response:
FAQ
Why do the tools use numeric IDs instead of string identifiers?
While the examples above show string IDs for clarity, the actual implementation uses numeric IDs. We found that when using string IDs (like domain names or slugs), language models were more likely to hallucinate plausible-looking but non-existent LLM.txt files. Using opaque numeric IDs encourages models to actually check the list of available files first rather than guessing at possible IDs.
Development
To run in development mode with automatic recompilation:
License
MIT
LLM.txt MCP Server
A Model Context Protocol (MCP) server that extracts and serves context from llm.txt files, enabling AI models to understand file structure, dependencies, and code relationships in development environments. This server provides comprehensive access to the LLM.txt Directory, supporting file listing, content retrieval, and advanced multi-query search capabilities.
Features
- Directory listing with local caching (24-hour cache)
- OS-specific cache locations:
- Multi-query search with context
Installation
Recommended: Using MCP Get
The easiest way to install is using MCP Get, which will automatically configure the server in Claude Desktop:
Manual Configuration
Alternatively, you can manually configure the server in your Claude Desktop configuration by adding this to your
claude_desktop_config.json:Tools
1. list_llm_txt
Lists all available LLM.txt files from the directory. Results are cached locally for 24 hours.
Example response:
2. get_llm_txt
Fetches content from an LLM.txt file by ID (obtained from list_llm_txt).
Parameters:
id: The numeric ID of the LLM.txt file
Example response:
3. search_llm_txt
Search for multiple substrings within an LLM.txt file.
Parameters:
id: The numeric ID of the LLM.txt file
queries: Array of strings to search for (case-insensitive)
context_lines(optional): Number of lines to show before and after matches (default: 2)
Example response:
FAQ
Why do the tools use numeric IDs instead of string identifiers?
While the examples above show string IDs for clarity, the actual implementation uses numeric IDs. We found that when using string IDs (like domain names or slugs), language models were more likely to hallucinate plausible-looking but non-existent LLM.txt files. Using opaque numeric IDs encourages models to actually check the list of available files first rather than guessing at possible IDs.
Development
To run in development mode with automatic recompilation:
License
MIT