io.github.woladi/macos-vision-mcp

Local OCR & image analysis via Apple Vision — no cloud, no API keys, ~97% fewer tokens on PDFs.