How to build document extraction pipelines with the munchr library. Use this skill whenever the user works with munchr, imports from 'munchr' or 'munchr/backends' or 'munchr/steps', wants to extract structured data or markdown from documents (PDFs, images, CSVs, HTML, XLSX, DOCX, emails), sets up OCR pipelines, or configures LLM-based extraction with schemas. Also trigger when you see the normalize/chunk/extract/merge pipeline pattern, or when someone asks about streaming structured JSON or markdown from documents.
2026-03-26