Sources

Notion

Ingest a Notion workspace export — every page, every database row.

The notion source ingests a workspace export from Notion in Markdown & CSV format. Each page becomes one item; database rows become items as well, with the database itself acting as a parent.

Get the export

In Notion, click Settings (top-left) → Settings → Workspace → Export content.
Choose Export format: Markdown & CSV.
Check Include subpages so nested pages come along.
Optionally check Include files & images (the export will be much larger).
Click Export. Notion emails a download link when ready.
Download the ZIP.

Register the export

bash

vault add-export \
  --source notion \
  --path ~/Downloads/notion-export.zip \
  --account workspace-name

Use your Notion workspace name as the account. If you export multiple workspaces (personal + a team), each gets its own --account.

What gets ingested

Item	Notes
One page item per `.md` file	`kind=page`, body is the Markdown, ts from filesystem mtime when in the export.
One row item per CSV row	`kind=database-row`, columns become structured metadata.
Page hierarchy	Parent / child relationships preserved as tags.
Database schema	Captured per-database from the CSV header.
Inline images	Image references in Markdown remain as relative paths into `staging/`.

Ingest it

bash

vault ingest --source notion

vault

$ vault ingest --source notion
Extracting notion-export.zip ...
Walking 1,402 pages and 18 databases ...
  pages:           1,402
  database rows:   8,210
Run 15 completed in 22.7s

Caveats

Image references are relative. Notion's export embeds images as Page%20Name%20<id>/image.png. The bytes live in staging/ after extraction; the Markdown body keeps the relative path. Search works on the page text, not the images.
Synced blocks and links to pages become Markdown links — the link target may resolve to another item in the export but the parser does not auto-follow at ingest time.
No comments. Page comments aren't part of the Markdown export.
Database views are not preserved — only the underlying rows. Views are a UI concept and don't survive the export.
For very large workspaces, prefer multiple smaller per-area exports over one massive ZIP — the parser handles large archives but extraction is bound by your local disk.