Initial setup: Knowledge base RAG system with LlamaIndex and ChromaDB

- Add Python project with uv package manager
- Implement LlamaIndex + ChromaDB RAG pipeline
- Add sentence-transformers for local embeddings (all-MiniLM-L6-v2)
- Create MCP server with semantic search, indexing, and stats tools
- Add Markdown chunker with heading/wikilink/frontmatter support
- Add Dockerfile and docker-compose.yaml for self-hosted deployment
- Include sample Obsidian vault files for testing
- Add .gitignore and .env.example

docker-compose.yaml

@@ -0,0 +1,32 @@
+version: "3.8"
+
+services:
+  knowledge-rag:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    container_name: knowledge-rag
+    volumes:
+      # Mount your obsidian vault here
+      - ${VAULT_PATH:-./knowledge}:/data/vault
+      # Persist ChromaDB vector store
+      - ./data/chroma_db:/data/chroma_db
+      # Persist embeddings cache
+      - ./data/embeddings_cache:/data/embeddings_cache
+    environment:
+      - VAULT_PATH=/data/vault
+      - EMBEDDING_MODEL=${EMBEDDING_MODEL:-all-MiniLM-L6-v2}
+      - EMBEDDINGS_CACHE_DIR=/data/embeddings_cache
+    restart: unless-stopped
+
+  # Optional: Watchtower for auto-updates
+  # watchtower:
+  #   image: containrr/watchtower
+  #   container_name: watchtower
+  #   volumes:
+  #     - /var/run/docker.sock:/var/run/docker.sock
+  #   environment:
+  #     - WATCHTOWER_CLEANUP=true
+  #     - WATCHTOWER_INCLUDE_STOPPED=true
+  #   command: --interval 3600 knowledge-rag
+  #   restart: unless-stopped