Skip to main content
Status: Available (Notion Desktop) Notion via notcrawl — it snapshots the Notion Desktop cache into a local SQLite archive (pages + blocks), which DuckBrain attaches read-only. The page is the document unit: its title plus its blocks’ text, concatenated in order.

Facet mapping

What gets ingested

  • Pages, not blocks — each live page becomes one record (title + its live block text). Empty / title-only pages are skipped.
  • Living documents — a page is stored once and updated in place when it changes (not a new copy per edit), so search always reflects the current version. Database-property values (a page’s column data) aren’t folded into searchable text yet, but a property edit still re-syncs the page.
  • Token-free — reads the local Notion Desktop cache; no NOTION_TOKEN or API access required (notcrawl also supports an API source if you want it).
  • Streaming. Notion Desktop writes its cache live, so the trigger lane re-runs notcrawl sync --source desktop + ingest when it changes — an edited page is searchable ~a minute later. The 6h schedule is the safety net.
Requires Notion Desktop installed and signed in on this Mac. Rows are classification: confidential and never leave the machine.