How does it handle OCR failures or blocked content?

The skill uses a robust fallback chain starting with segment-based PDF uploads, moving to per-page image OCR, and finally native PDF text extraction. It also provides specific workflows for handling 'contentFilter' blocks by splitting segments.

How do I know if the OCR process was successful?

The workflow requires running verify_ocr.py and audit_ocr_integrity.py. A batch is only considered complete when these checks are clean and no failure JSON files remain in the output directory.

Can it handle mathematical formulas and LaTeX?

Yes, it includes specialized scripts like repair_math_delimiters.py to clean up OCR-generated math and ensure LaTeX delimiters are correctly formatted.

What file formats does GLM-OCR support?

GLM-OCR supports PDF, PPT, PPTX, and standard image formats like JPG and PNG for conversion to Markdown.

GLM-OCR Markdown Toolkit

Name: GLM-OCR Markdown Toolkit
Author: sunrisever

bysunrisever

•

Data Science & ML

Converts PDFs, PPTs, and images into clean Markdown using AI-assisted GLM OCR with strict integrity verification.

The GLM-OCR skill provides a professional-grade pipeline for digitizing complex documents such as textbooks, lecture slides, and screenshots into structured Markdown. It utilizes Zhipu GLM-OCR for high accuracy and implements a rigorous verification system to prevent silent failures and data loss. With specialized handlers for LaTeX math delimiters, automated image auditing, and multi-stage fallbacks (Segment PDF -> Per-page image -> Native text), it ensures that even challenging scanned materials are accurately converted for use in digital knowledge vaults and documentation systems.

Key Features

012 GitHub stars

02Multi-stage fallback OCR for PDF, PPT, and PPTX files

03Duplicate image detection and similarity auditing tools

04Automated LaTeX math delimiter repair and Markdown cleanup

05Reference book metadata and directory page generation

06Strict integrity auditing and verification scripts to prevent data loss

Use Cases

01Converting scanned textbooks and academic papers into searchable Markdown

02Transforming presentation decks into structured study guides or documentation

03Auditing and repairing corrupted or incomplete legacy OCR outputs

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add sunrisever/glm-ocr glm-ocr

For use in Claude.ai and ChatGPT