MCP Qdrant Docs

Example Use Cases

• 1

  Search and answer questions over scraped website documentation using a Qdrant vector index.

• 2

  Expose a site-specific QA tool through an MCP interface (e.g., Claude Desktop) runnable via npx.

• 3

  Build custom knowledge bases by crawling, chunking, embedding, and indexing web content for retrieval-augmented generation.

Description

This server scrapes a target website, chunks and embeds the scraped text with a sentence-transformer model, and stores embeddings and content chunks in a Qdrant collection. It then exposes a dynamically named MCP tool (e.g., ask_reactrouter_docs) that embeds user queries, searches Qdrant for relevant chunks, and returns the matching content. It is designed to be run via npx or installed globally and integrates easily with MCP clients like Claude Desktop. Configuration is via CLI options or environment variables and supports forced reindexing, collection naming, and model selection.