Document Library MCP

Example Use Cases

• 1

  Ingest and semantically search across markdown, code, OpenAPI, and HTML files to surface the most relevant content chunks.

• 2

  Build RAG contexts by retrieving top-k relevant chunks filtered by type or tag and combining them into prompts.

• 3

  Lightweight note management and summaries with access via a note:// URI scheme and programmatic tools to add, list, and summarize notes.

Description

This server provides an in-memory note store with a custom note:// URI scheme, tools to ingest and robustly chunk Markdown, Python, OpenAPI, and HTML content, and vector-based semantic search over the resulting chunks. It includes a smart ingestion path that uses Gemini to extract technical content and a mistune-based chunker to preserve code and narrative as separate chunks. The service also provides utilities for tagging, metadata updates, deletion, and prompts (e.g., summarizing notes) to help agents construct context for retrieval-augmented generation (RAG). Configuration hooks for Ollama models and a PostgreSQL-backed configuration are documented for integration.