Overview
What is プロジェクトディレクトリ概要?
プロジェクトディレクトリ概要 is a project that provides MCP (Model Context Protocol) servers to reference API documentation. It uses the reference functionality and includes scripts for crawling web documentation and serving it via MCP, targeting users who want to access API docs through Claude.
How to use プロジェクトディレクトリ概要?
Clone the repository, install uv, create a virtual environment, and install dependencies (mcp[cli] and PyYAML). Prepare documentation by crawling (or downloading for OpenAPI), then run one of the MCP server scripts (e.g., uv run gemini_docs.py). Finally, update claude_desktop_config.json with the appropriate command and restart Claude.
Key features of プロジェクトディレクトリ概要
- Provides MCP servers for Gemini, Anthropic, and OpenAPI documentation.
- Crawls web documentation using a customizable script (
crawl_all.py). - Supports presets for Gemini and Anthropic documentation.
- Allows custom crawling with options like base URL, selector, and delay.
- Integrates with Claude Desktop via
claude_desktop_config.json.
Use cases of プロジェクトディレクトリ概要
- Referencing Gemini API documentation through Claude using MCP.
- Accessing Anthropic API documentation via the MCP server.
- Serving OpenAPI documentation (downloaded from GitHub) through MCP.
- Crawling and serving custom documentation sites for use with Claude.
FAQ from プロジェクトディレクトリ概要
What documentation sources are supported?
The project supports Gemini, Anthropic, and OpenAPI documentation. Gemini and Anthropic docs are crawled; OpenAPI docs are downloaded from the OpenAI OpenAPI GitHub repository.
How do I set up the MCP server?
Install uv, create a virtual environment, install dependencies, run the desired Python script (e.g., gemini_docs.py), and configure claude_desktop_config.json with the command pointing to the script.
How do I crawl documentation?
Use crawl_all.py with a preset (--preset gemini or --preset anthropic) or custom options like --base-url, --start-path, --output_file, --selector, --delay, --max-pages, and --use-selenium for dynamic pages.
What are the runtime dependencies?
Python, uv, and the packages mcp[cli] and PyYAML. For crawling, additional options like Selenium may be needed for dynamic pages.
Where does the crawled data live?
Crawled documentation is saved in the document/ directory as JSON files (for Gemini and Anthropic) or as a YAML file (for OpenAPI).
