Tesseract Ocr · dsh plugin
maxwell-feng/dsh-tesseract-ocr · model-adapters
dsh plugin: OCR attached images locally with Tesseract — text-only models can see, privacy-first.
★ 5GitHub stars
1Forks
2026-09-04Last updated
JavaScriptLanguage
MITLicense
Key features
- Privacy defaultimage bytes are OCR'd locally and not sent to the provider.
- Set passthroughtrue only if you intentionally want genuine vision models to receive original image bytes.
- Works with any provider/model in dsh
- by default every attached image is OCR'd before the request leaves the machine.
- Vision-model passthrough is opt-in (passthrough: true).
Requirements
- Language packstesseract-ocr-eng (usually pulled in by the base package), tesseract-ocr-chi-sim, tesseract-ocr-chi-tra, tesseract-ocr-jpn, … The language config joins multiple tags with +, e.g.
- Recognition quality depends on the installed language packs and psm
Install command
dsh plugin --profile web add github:maxwell-feng/dsh-tesseract-ocr