Kuzco Image Text Extraction
Extracts text from images, documents, and screenshots using local vision language models in R.
Integrates local vision language models into R workflows to perform optical character recognition (OCR) on screenshots, scanned documents, signs, and annotated charts. Powered by the kuzco package and ellmer tool framework, this skill enables Claude Code to transcribe text directly from local JPEG and PNG files with configurable prompts, target language specification, and confidence scoring for transparent, faithful text extraction.
主な機能
01Extracts text from screenshots, scans, labels, and charts using local vision LLMs
02Provides compact JSON output with transcription confidence metrics
03Integrates seamlessly with the R ellmer and kuzco tool ecosystem
04Runs locally via Ollama to ensure complete data privacy without cloud APIs
05Supports custom prompt instructions and target language specification
0669 GitHub stars
ユースケース
01Extracting code snippets, terminal logs, or error text directly from screenshots
02Digitizing text from photographed documents, receipts, or product labels
03Reading annotations, axis labels, and captions from scientific charts and diagrams