pdf-reader - SKILL.md Agent Skill

name: pdf-reader description: Reads PDF, DOCX, and image text (OCR) and can pinpoint schedule lines from syllabus documents. Use when the user asks to extract text from docs/images or analyze deadlines and class schedules.

PDF + OCR Reader Skill

Use this skill for document extraction and schedule pinpoints.

What it supports

PDF text extraction (pdftotext)
DOCX text extraction (word/document.xml)
Image OCR (tesseract if available, otherwise macOS Vision OCR via Swift)
Syllabus schedule/deadline pinpoint analysis

Commands

# Legacy PDF-only command
node "{baseDir}/scripts/read_pdf.mjs" --file "/path/to/your/document.pdf"

# Unified reader (PDF/DOCX/Image) + optional schedule analysis
python3 "{baseDir}/scripts/read_docs.py" \
  --path "$HOME/Desktop/syllabus" \
  --analyze-schedule \
  --save "$HOME/Desktop/syllabus/syllabus_schedule_report.md" \
  --json-out "$HOME/Desktop/syllabus/syllabus_schedule_report.json"

# Shortcut wrapper for syllabus folder
bash "$HOME/openclaw_pro/workspace/scripts/syllabus_pinpoint.sh" \
  "$HOME/Desktop/syllabus"

Output contract

ANALYSIS_STATUS=ok|error
FILES_SCANNED=...
EVENTS_FOUND=...
REPORT_MD=...
REPORT_JSON=...
PINPOINT_TOP_START ... PINPOINT_TOP_END

Notes

OCR quality is highest on clear images (300dpi+).
For scanned PDFs, OCR fallback uses pdftoppm + OCR.
If OCR backend is unavailable, script still returns extracted text from readable PDFs/DOCX.