dsh-simpletex-ocr

In the DSH Web GUI conversation flow, users often want to turn a screenshot of a math formula into editable LaTeX, or extract a page of product documentation into Markdown text. Manually assembling SimpleTex API parameters, handling authentication and callbacks is both tedious and error-prone.

dsh-simpletex-ocr collapses the entire pipeline into a single conversation tool called simpletex_recognize, offering three modes: formula, text, and auto. In auto mode, the plugin decides whether to treat the image as a formula or plain text based on contextual keywords. The input box supports drag-and-drop, paste, and file picker; after sending, a clickable thumbnail appears inline and the model reads the reference to complete recognition. The API key is written to the local credential service via a GUI configuration card, never displayed or stored in a plaintext settings file. Images are validated for type and size before upload, with clear error messages for oversized or non-image files.

Ideal for developers and researchers who frequently handle math formula screenshots or technical documentation OCR, especially those who have already built workflows in DSH Web GUI and want to keep recognition steps inline within the conversation rather than switching to a browser or a script.