your-papa/obsidian-Smart2Brain

Implements a RAG (Retrieval-Augmented Generation) pipeline that embeds vault notes as vectors and retrieves contextually relevant content to ground LLM responses, with automatic reference linking back to source notes. Supports both local inference via Ollama and cloud-based models like ChatGPT, using Langchain and Orama vector store with hierarchical tree summarization for note processing. Persists conversations and allows flexible switching between different LLMs for specialized tasks.

1,002 stars. Actively maintained with 83 commits in the last 30 days.