Add Knowledge Base Hybrid RAG MCP Server

[hwandam77]

Summary

Add Knowledge Base Hybrid RAG MCP Server - a powerful 37-tool MCP server for Hybrid RAG with PostgreSQL/pgvector.

Features

Feature	Description
Hybrid Search	Semantic + Keyword search (BM25 + Vector) with configurable weights
LLM Reranking	+5-10% accuracy improvement using LLM relevance scoring
In-Memory Caching	2-3x speed improvement with TTL-based cache
Knowledge Graph	Entity/relation-based knowledge management (6 tools)
Auto Entity Extraction	LLM-based NER with automatic KG save
Conversational Memory	Session-based memory for conversational RAG
Multiple Chunking	fixed, paragraph, semantic strategies
Evaluation Metrics	Precision, Recall, F1, MRR for quality measurement

Tool Categories (37 total)

• System: 1 tool (health check)
• Search: 5 tools (hybrid, semantic, keyword)
• AI Features: 5 tools (Q&A, summarize, tagging)
• Document CRUD: 7 tools
• Knowledge Graph: 6 tools
• Unified Search: 1 tool
• Reranking: 2 tools (v2.1.0)
• Caching: 2 tools (v2.1.0)
• Entity Extraction: 2 tools (v2.1.0)
• Memory/Session: 3 tools (v2.1.0)
• Chunking: 1 tool (v2.1.0)
• Evaluation: 2 tools (v2.1.0)

Requirements

• PostgreSQL with pgvector extension
• Ollama with embedding model (qwen3-embedding:0.6b)
• Ollama with LLM model (gemma3:12b or similar)

Repository

• Source: https://github.com/hwandam77/Knowledge-Base-Hybrid-RAG
• License: MIT
• Docker Image: Dockerfile included

Checklist

• MIT License
• Dockerfile in repository
• server.yaml with all required fields
• tools.json with 37 tools
• readme.md with documentation

---

Thank you for reviewing! 🙏