/docs/migrations/

Migrating from existing knowledge systems

You don't have to rewrite your knowledge to bring it into OKF. The OKF Knowledge Hub imports exports from four common systems – Notion, Obsidian, Logseq, and Confluence – and turns them into individual OKF concepts. This page shows the export for each, states what the importer keeps, and is honest about what it cannot do.

The path is the same everywhere: in the upload dialog choose "Import from another service" and upload the ZIP. The source is usually detected automatically but can be set manually. An import counts toward the monthly limit like any upload.

Notion

Export the required area as Markdown & CSV and upload the resulting ZIP. Markdown and HTML pages become concepts. Rows from a CSV database each become a concept of type Collection Item – so a Notion database with 40 rows yields 40 concepts.

Tip: Notion appends long hash IDs to file and folder names. The importer handles them, but clean page titles in Notion lead to cleaner titles in OKF.

Obsidian

Zip the vault folder. The importer processes Markdown files, YAML metadata, tags, and wiki links. The technical .obsidian folder is ignored. If you already use frontmatter, you have an edge here: existing tags and simple metadata move straight into the OKF concepts.

Logseq

Export the graph as Markdown and upload a ZIP containing pages/ and, where present, journals/. Journal files are recognized as Journal Entry. One limit of the current version: simple block references are not carried over as interactive references.

Confluence

Export the space as HTML and upload the ZIP. The importer converts headings, paragraphs, lists, links, code blocks, and emphasis to Markdown. Confluence labels become tags where detected. Complex layouts, macros, tables, or embedded apps may be simplified – plan a visual check.

What the importer keeps

Across all four systems the import keeps: page titles and text, detectable tags or labels, the folder structure as OKF paths and categories, local Markdown links and wiki links where the target is unambiguous, existing simple metadata such as type or source URL, and an auto-generated short description from the first suitable paragraph.

Each concept gets a type its source system suggests: Notion pages become Knowledge Page, Confluence pages Documentation Page, Obsidian and Logseq notes Note. The importer recognizes Logseq journals as Journal Entry, Confluence blogs as Blog Post, and each row of a Notion database as a Collection Item. Any type from the source frontmatter takes precedence.

After import, the Hub returns a report: documents discovered, concepts created, documents skipped, links rewritten, and broken links. Then spot-check structure and content.

Where the limits are

Honesty saves frustration. The importer targets text-based content. Images, PDFs, and other attachments are not currently stored as separate content. Complex Confluence constructs may be simplified. Unresolvable internal links end up as broken links in the result – often a sign of renamed pages, duplicate titles, or links to content that was not exported. And if the source is not detected automatically, select it manually in the dialog.

Cleaning up after import

An import gets you 80 percent of the way; the last 20 percent decide search and retrieval quality. Three moves almost always pay off.

Beschreibungen schärfen. Die automatisch erzeugte Beschreibung ist der erste Absatz, nicht immer die beste Zusammenfassung. Ersetzen Sie sie durch einen Satz, der das Konzept für sich verständlich macht. / Sharpen descriptions. The auto-generated description is the first paragraph, not always the best summary. Replace it with one sentence that makes the concept understandable on its own.
Tags vereinheitlichen. Über verschiedene Quellen hinweg entstehen leicht SEO, seo und Seo nebeneinander. Ein einheitliches Schema verbindet Konzepte im Graphen, die sonst isoliert bleiben. / Unify tags. Across sources, SEO, seo, and Seo easily pile up side by side. One consistent scheme connects concepts in the graph that would otherwise stay isolated.
Kaputte Links prüfen. Nicht jeder kaputte Link ist ein Fehler – in OKF darf ein Link auf noch nicht geschriebenes Wissen zeigen. Aber prüfen Sie die Liste, viele sind echte Tippfehler oder umbenannte Ziele. / Check broken links. Not every broken link is a fault – in OKF a link may point to not-yet-written knowledge. But review the list; many are real typos or renamed targets.

Editing happens directly in the Concept Library reader: change title, description, type, category, tags, and body, save, done. The semantic index rebuilds afterward.

Continue

What a clean target schema looks like is on the frontmatter schema page. To automate recurring imports, see CLI and CI/CD. For import warnings, troubleshooting helps.

Import-Verhalten und Grenzen nach der Anleitung des OKF Knowledge Hub, Abschnitt Import. / Import behavior and limits per the OKF Knowledge Hub guide, import section.