Overview
What is Docs To Llm?
Docs To Llm is an MCP server that processes technical documentation into llm.txt format. It generates two files — a short version with titles and links, and a full version with complete content — for use as context in LLM environments like Cursor, Windsurf, Cline, or Roo Code.
How to use Docs To Llm?
Clone the repository and install dependencies with uv pip install -r requirements.txt or pip install -r requirements.txt. Configure the server in your MCP settings (e.g., mcp_settings.json) with the command python and the path to src/server.py. Invoke the process_documentation tool to crawl a documentation URL, providing the URL, library name, and output directory. The server returns paths to the generated llm_{library_name}_short.txt and llm_{library_name}_full.txt files.
Key features of Docs To Llm
- Automatic detection of documentation navigation sections
- Conversion of relative URLs to absolute URLs
- Removal of unnecessary HTML elements (scripts, styles, menus)
- Progress reporting during processing
- Detailed error logging
- Smart scoring system to find relevant documentation links
- Fallback mechanisms when automatic detection fails
- Sanitized filenames based on library names
Use cases of Docs To Llm
- Generate concise documentation context for LLM-powered code assistants like Cursor or Windsurf
- Create a full-text local copy of a library's docs for offline analysis
- Produce short link-only summaries of large documentation sites
- Integrate documentation ingestion into automated build or tooling pipelines
- Feed structured documentation into Roo Code or Cline for context-aware prompts
FAQ from Docs To Llm
What Python version is required?
Python 3.7 or higher is required.
How do I install the dependencies?
Use uv pip install -r requirements.txt (recommended) or pip install -r requirements.txt.
Where are the output files saved?
You specify the output directory via the txt_save_path parameter in the process_documentation tool call. The server creates two files there.
How do I configure Docs To Llm in Cline or Roo Code?
Add a docs-to-llm entry to your mcp_settings.json with command python and args pointing to the full path of src/server.py.
What tool does the server expose?
The server provides one tool: process_documentation, which accepts a URL, library name, and save path, and returns success status and file paths.
