- Add 'summary' vectorized field to Chunk collection description - Update vectorization strategy (text/summary/keywords) - Add HNSW + RQ vector index configuration section - Correct Docker config: BGE-M3 ONNX is CPU-only (not CUDA) - Add llm_summarizer.py and summary generation scripts to project structure - Update annexe with accurate GPU/VRAM information - Remove incorrect GPU configuration example 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>