Corpus Ingestion — MarkItDown Pipeline

Purpose

This skill governs how raw source materials are converted into markdown and promoted into a curated corpus. MarkItDown is the conversion layer, not the intelligence layer. It normalizes heterogeneous formats into a common markdown baseline so that later skills can clean, annotate, split, analyze, and retrieve from that content.

Use this skill when:

• Ingesting an author's books, essays, or published works
• Adding sermon transcripts, lecture recordings, or audio content
• Processing slide decks, workbooks, or PDF handouts
• Pulling web pages, articles, or archive materials into a corpus
• Preparing any source material for theme/voice analysis, RAG, or course generation

Do not skip this skill when adding new source material — raw files should always pass through the ingestion pipeline before entering curated storage. Skipping tiers creates data provenance problems.

Three-Tier Pipeline Architecture