Personalized Document Assistant

Example Use Cases

• 1

  Search and retrieve information from a user’s local document collection across PDF, Word, Excel, Markdown, and text files.

• 2

  An MCP-compatible backend to provide semantic document QA and context for conversational assistants.

• 3

  Index and query documents using custom embedding models and a local vector store for private, on-premise data access.

Description

This MCP server loads documents from specified local paths, builds vector embeddings, and exposes semantic search and read interfaces via the MCP protocol. It supports many common formats (PDF, TXT, CSV, Excel, Word, Markdown, etc.) and uses a vector database for retrieval. The server provides standard MCP methods (list_resources, read_resource, list_tools, call_tool) and a query_document tool for asking questions about indexed content. Initial indexing may take time and query performance depends on document volume and the chosen embedding model.