OCR-level vision bridge for DeepSeek Harness: paste images, read them locally (macOS Vision / Tesseract), answer with your text-only DeepSeek model
dsh plugin --profile web add dsh-ocr-bridge
View the source on GitHub