Integrates image Q&A, OCR, grounding, cropping, tracing, pixel comparison, HTML screenshots, and a task Skill into DSH Web and Headless. Configure a vision provider; first use may set up Python.
ocr
Plugins tagged “ocr” in the DSH plugin directory. Check individual sources and trust before installation.
Parses local PDF, Office, Markdown, HTML, CSV, and text files, with offline OCR; reads only the session workspace or allowed roots, with a bundled OCR runtime for Windows x64.
Adds vision routing and image tools for text-only DSH agents, including OCR, grounding, cropping, and pixel comparison; the default cloud chain sends images remotely unless Local-only vision is enabled.
Reads pasted images for text-only models and returns OCR, layout, and semantic evidence; a vision engine must be available.