Second Brain — Ingest

Process raw source documents into structured, interlinked wiki pages. Two tiers exist: light (fast, batch-safe source page) and deep (full entity/concept extraction).

Collect & Enrich Clippings (runs first)

If the vault root has a Clippings/ folder (the Obsidian Web Clipper's default save target), sweep it before identifying sources — the user should never have to move clips into raw/ by hand.

For each Clippings/*.md: