Commit History

  1. Speed up CPU inference: halve token limits, pre-download models, fix OMP threads⁠

    cgoodmaker +1: Claude Opus 4.6
    4af4003

  1. Use bfloat16 on CPU to halve memory (8GB vs 16GB float32)⁠

    cgoodmaker +1: Claude Opus 4.6
    0989643

  1. Fix MCP subprocess deadlock: use stderr=None instead of PIPE⁠

    cgoodmaker +1: Claude Opus 4.6
    da343a7
  2. Add timeout and stderr logging to MCP subprocess to debug tool hangs⁠

    cgoodmaker +1: Claude Opus 4.6
    c376e14
  3. Remove unused files: old Gradio frontend, dead model code, test artifacts⁠

    cgoodmaker +1: Claude Opus 4.6
    672ed11
  4. Force MCP tool models to CPU to avoid GPU VRAM contention with MedGemma⁠

    cgoodmaker +1: Claude Opus 4.6
    1a97904
  5. Add missing deps: sentence-transformers, pdfplumber, scipy for MCP tools⁠

    cgoodmaker +1: Claude Opus 4.6
    5157ba3
  6. Add RAG Phase 4 management guidance, rebuild guidelines index (286 chunks), post-analysis hint UI⁠

    cgoodmaker +1: Claude Opus 4.6
    5241b71
  7. Use dtype instead of deprecated torch_dtype in model_kwargs⁠

    cgoodmaker +1: Claude Opus 4.6
    82f82ac
  8. Redesign chat UI and fix MedGemma generation config issues⁠

    cgoodmaker +1: Claude Opus 4.6
    58a4476

  1. Fix requirements: opencv-headless (no libGL needed), remove unused gradio, add faiss-cpu and huggingface_hub

    4cbee96
  2. Copy mcp_server/ into Docker image

    dfb4396
  3. Include mcp_server/ in Docker image — required at runtime for tool calls

    2068e4f
  4. Pass HF_TOKEN explicitly to pipeline() for gated model auth

    b08f876
  5. Use HF_TOKEN env var to authenticate for gated MedGemma model

    bb7e939
  6. Fix Dockerfile: install curl before using it for NodeSource setup

    076039f
  7. Migrate FAISS index to Git LFS

    b419917
  8. Track FAISS index via Git LFS

    6bc45eb
  9. Add HF Spaces Dockerfile, Git LFS for model weights

    72b1012
  10. Initial commit — SkinProAI dermoscopic analysis platform⁠

    86f402d