docs/memory/PROGRESS.md

Progress: Cora

[COMPLETED]

• Unified FastAPI/Gradio entry point (app.py).
• CoraEngine with multi-model fallback.
• CoraVision with CLIP embeddings and YOLO tagging.
• CoraMemory with ChromaDB persistence.
• Etymology API integration endpoints.
• Hugging Face Spaces deployment compatibility (HF_TOKEN fallback).
• Verified valid token permissions.
• Documentation Stack initialized and moved to docs/memory/.
• Generation pipeline hardened with proactive image loading (.load()) and model fallback.
• Added AirLLM and Ollama backend support to CoraCurator.
• Updated app.py for dynamic backend reporting.
• Fixed persistent image reading errors and added truncated image support.
• Enhanced historical fidelity with social realist visual tokens and ethnographic persona.
• Successfully deployed to Hugging Face Spaces.

[CURRENT FOCUS]

• Confirm stability in Hugging Face Space.
• Improve RAG fallback hit rate by enriching museum loaders.

[TECHNICAL DEBT]

• Loader scripts currently have limited error handling for network timeouts.
• Background archival task in app.py uses basic threading.Thread; consider a more robust queue if traffic increases.