Personal RAG MCP Server

# RAG Pipeline Configuration retrieval: strategy: vector # Options: vector, hybrid (future) top_k: 5 score_threshold: 0.0 chunking: chunk_size: 512 # tokens overlap: 50 # tokens encoding: cl100k_base # GPT-4 tokenizer reranking: enabled: false # Future: cross-encoder reranking model: null query_expansion: enabled: false # Future: LLM-based query expansion method: null