Run3 document content extraction
[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.
npx -y skills add cxcscmu/SkillLearnBench --skill run3_document_content_extractionAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
Robustly extracts text from PDF, DOCX, and PPTX files while maintaining structural integrity for multi-column layouts and nested elements.
SKILL.md
0.8 KB, as published. Nobody here has run it
To ensure high-fidelity text extraction:
- PDFs: Utilize
pdfplumberorPyMuPDFwith layout-aware parsing enabled to prevent text jumbling in multi-column academic papers. - DOCX: Use
python-docxto extract text from body paragraphs, tables, and headers while ignoring metadata that may cause noise. - PPTX: Use
python-pptxto iterate through slides, specifically extracting text from text frames and shapes to capture slide content. - Fallback: If text extraction fails (e.g., scanned PDFs), use an OCR fallback (e.g.,
pytesseract) to ensure the file is still categorizable.