Docs
Createdarabold
Indexes and searches third-party library documentation using semantic splitting and vector embeddings.
About
Provides a Model Context Protocol (MCP) server for fetching, processing, and searching documentation for software libraries and packages. It scrapes content, splits it into chunks using semantic techniques, generates vector embeddings with OpenAI, and stores the data in SQLite, leveraging `sqlite-vec` and FTS5 for hybrid search. The server supports versioning, job management, and includes a CLI tool for local interaction.
Key Features
- Includes a CLI tool for local management
- Offers job management for scraping tasks (start, status, cancel)
- Supports versioning of documentation
- Scrapes and indexes documentation from URLs
- 4 GitHub stars
- Utilizes semantic splitting and OpenAI embeddings for indexing
Use Cases
- Integrating library documentation into Claude Desktop
- Managing documentation scraping jobs
- Searching documentation for specific library versions
