This skill should be used when the user asks to "OCR these scans", "extract text from this archive", "track what my extraction run did", "rerun the bad pages", "compare two runs", or otherwise wants document extraction with provenance, quality review, cost budgets, or reproducibility — digitization projects, archival scans, batch PDF extraction, OCR auditing. Also use it when operating or extending the pageledger CLI or writing a PageLedger adapter.
2026-07-17