Skip to content
Recued
Menu
← Back to packs

Tesseract OCR Pack

by recued-core v1 app_pack

V3 deterministic image-OCR capability pack. By-value connector composition (service_kind=cli, no separate ingredient): one local tesseract CLI ingredient exposes bounded operations for readiness, installed language discovery, plain-text OCR, and TSV OCR with word boxes/confidence. Fixed-function and reproducible: OCR happens locally before any model sees the image, no caller-supplied Tesseract config files, no user words/patterns, no tessdata path override, no arbitrary flags, no stdin, and no shell wrapper. OCR outputs are faithful local file writes into engine-managed throwaway temp dirs, captured as result.file_ref, and consumed by downstream ai-* steps through Gateway-gated file refs. The source can be a data.file ref or a trusted local path. Requires the tesseract binary and requested language data on PATH (Debian/Ubuntu: apt install tesseract-ocr tesseract-ocr-eng; macOS: brew install tesseract). Local and no-egress, but image decoding still depends on Leptonica/Tesseract format parsers; prefer file_ref sources and keep the OCR stack patched.

What this pack installs

  • tesseract Ingredient

Trust & control

What installing this whole pack would let it do. Recued grants these permissions at install — review them there before approving.

Where it runs

Server only Runs on your server.

What it's allowed to do

Write Local CLI tesseract external change
Read: tesseract.version · No approvalRead: tesseract.list_languages · No approvalWrite: image.ocr · No approvalWrite: image.ocr_tsv · No approval

Data it touches

SurfacesLocal CLI

About

Tags

pack:tesseracttesseractcliocrimagetexttsvlanguage-discoveryparsedeterministiclocal-servicev3composition