Imports: bring your notes in

Import Notion pages and databases with akb-importer. For activity data such as Netflix viewing history, see data ingestion.

Crawl Notion

The importer reads Notion through its API, not an HTML or Markdown export: the API keeps relations, typed values and other data that exports drop. Create an integration, share your pages with it (read access is enough; sharing a top-level page shares everything below it) and crawl everything it can see into a local directory:

cargo build --release -p alexandria-importer
export NOTION_TOKEN=ntn_...
./target/release/akb-importer notion-crawl /path/to/notion-crawl

To crawl only some pages, pass each with --root (an ID or URL). The crawl follows subpages, databases with all their rows and every database reached through a relation, and downloads Notion-hosted files. It is a self-contained backup. If it is interrupted, or reports pages the integration cannot read, share them and rerun with --resume.

Preview

./target/release/akb-importer notion \
  /path/to/notion-crawl /path/to/alexandria/workspace \
  --dry-run --verbose

The preview does not contact Notion and does not initialize or modify the destination. It reports counts, source/destination IDs, parents and warnings.

Import

Back up your destination and close apps that could edit it during migration. Run the same command without --dry-run:

./target/release/akb-importer notion \
  /path/to/notion-crawl /path/to/alexandria/workspace

Open the workspace and review representative notes, database rows, toggles, properties, media and parent/child placement before trusting the migration.

What is preserved

Pages become notes; databases become native tables/tags with typed properties: text, numbers, currencies as Money, percentages, selects and statuses, dates with offsets and ranges, people, checkboxes, URLs/emails/phones, files, places, created/edited times and relations. Two-way relations become native pairs. Toggles remain expandable; links to crawled pages and blocks target stable note IDs and anchors. Crawled files become content-reused media; external media is downloaded during the import.

Reimport matches Notion IDs, never titles. Duplicate titles do not collide. It replaces source-owned content, properties and placement while retaining unrelated memberships, personal columns/children, original creation timestamps and pages absent from the new crawl. It never automatically deletes stale pages. An import is not transactional; keep your backup and rerun the same crawl after a filesystem failure.

Limits and report

Each import writes import-report-<timestamp>.csv beside the vault, listing what it could not reproduce: links to pages outside the crawl (kept as Notion URLs), blocks the API cannot return such as Google Drive (placeholders linking to Notion), unreachable relation targets, unsupported column types and linked images or files that could not be downloaded (their original links are kept; importing again retries them).

Formula and rollup results, comments, permissions, history, database views and column order (the API does not expose it; columns are ordered by name) are not reconstructed. Embeds other than YouTube stay inert placeholders.

See alexandria-importer/README.md for every option.