Coming from Logseq
Everything the importer does, in detail
Logseq is moving to a database. Cairn keeps your notes as plain Markdown files you own, and it can bring your Logseq graph along.
Asking an AI whether your graph will survive the move? Point it at the plain-text version: cairn-notes.com/logseq.md
The short version
- Cairn imports Logseq Markdown file graphs on the Mac: File → Import from Logseq…
- Your Logseq folder is never touched. Cairn reads your graph and writes copies into Cairn. Nothing in your Logseq folder is moved, changed or deleted. Your graph stays exactly where it was, and you can keep using Logseq.
- Nothing disappears silently. Everything either converts, or is kept and clearly marked. When the import finishes, a report lists every construct that wasn’t fully converted, by category.
An honest note
Importers for block-based Markdown outliners are notoriously rickety. We’ve worked hard on this one, and we think it’s better than most. But we built and tested it against real graphs: our own, plus a set of public Logseq graphs. Yours is different. You may use properties differently than we do, or rely on something (like Logseq’s time-tracking logs) that we treat as clutter.
So give it a try. If something comes across wrong, email support@cairn-notes.com with what you expected and what you got. We’re genuinely open to adjusting the importer for how you work.
Try it safely
- The importer only reads your Logseq folder.
- Before importing, Cairn offers to back up your Cairn notes.
- Want a dry run first? Import our sample graph, a fictional book project.
What converts
| In Logseq | In Cairn |
|---|---|
| Pages and journals | Pages and daily journal pages |
[[Page links]] | [[Page links]]. A link to a page that doesn’t exist creates that page, so nothing dangles. |
Namespaced pages (a/b/c) | A page titled a/b/c |
alias:: | Links to any alias still reach the page |
#tag, #[[tag]], tags:: | Cairn tags |
TODO, LATER | To-do checkbox |
DOING, NOW | In-progress checkbox |
DONE | Done checkbox |
CANCELED | Canceled checkbox |
WAITING, IN-PROGRESS | To-do checkbox plus a #logseq/waiting or #logseq/in-progress tag |
Priorities [#A] | A #priority-a tag |
Your own properties (author:: Ada) | Kept exactly; shown as Author: Ada |
template:: | A #template tag |
| Numbered lists | Numbered lists (starting at 1) |
| Images and other assets | Copied into Cairn alongside the page |
{{embed ((block))}} | A live embed (see below) |
{{embed [[Page]]}} | A [[Page]] link |
What’s kept and marked
These come across with their original text intact, plus a visible marker saying what they were:
- Queries (
{{query …}}): kept, but not executed. Cairn has no query engine. - Whiteboards and drawings: the file is copied, and a placeholder page points to it.
- Admonitions (
#+BEGIN_NOTE…#+END_NOTE): become marked blockquotes. - Plugin and unknown macros (
{{renderer …}}and similar): kept as text and marked. - References whose target isn’t in the graph: kept as written and marked.
How Cairn works differently
Stable block IDs
Logseq’s block UUIDs don’t carry over. Cairn gives every block its own stable ID; those IDs are what Cairn’s built-in MCP server uses when an AI assistant works with your notes. Every reference in your graph is rewired to the new IDs during import.
Embeds are “projections”
{{embed ((block))}} becomes a live Cairn embed. We call these projections. They show the original block’s current content and stay live across your devices.
No inline references
Every reference becomes its own block or a link, and you choose how. Cairn embeds are block-level: one embed is one block, and there’s no embedding mid-sentence. So a bare ((block)) reference in the middle of a sentence is handled by one import setting:
- Convert to embeds (default): the block is split into three: the text before, the embed as its own block, and the text after. Your words stay; the reference becomes live.
- Replace with text: the referenced block’s text is pasted in place as a one-time snapshot. No split, but it won’t update.
If a referenced block can’t be linked live, Cairn falls back to a snapshot of its text; it never leaves a raw ID behind.
Labeled references become links
[some text](((block))) becomes a link showing “some text”. Clicking it opens that page and scrolls to the block. (It’s a link in Logseq too, so it stays a link.) This ignores the setting above.
Properties work like Roam attributes
status:: drafting shows as Status: drafting. Your properties are kept exactly as you wrote them, and a property’s name links to its own page, so every use of Status:: collects in one place. Page-level properties sit at the top of the page. title:: is recorded but not applied: a page’s title comes from its file name, so existing links keep working.
Logseq’s own bookkeeping
Logseq quietly stores data for its own features: :LOGBOOK: time-tracking entries, flashcard scheduling, PDF highlight positions, collapsed/expanded state, and similar. A second import setting decides what happens to it:
- Clean (default): drops that machine-generated data. The report says how much was dropped.
- Keep everything: keeps all of it, marked, exactly as written.
In both modes, your own properties and any drawer Cairn doesn’t recognize are always kept. If you rely on your clock history or flashcard data, choose Keep everything.
What isn’t supported (yet)
- Org-mode (
.org) files aren’t imported; only Markdown (.md) pages and journals are. - Logseq’s new database format isn’t supported. Cairn imports the classic file-based (Markdown) graphs.
- Import runs only on the Mac. Your notes then sync to your iPhone through iCloud.
If you need any of these, tell us. It helps us decide what to build next.
Coming from Obsidian or Roam Research?
Importers for both are on our roadmap, and Obsidian is likely first. Want early access? Email support@cairn-notes.com. Hearing from you moves it up the list.
Questions or problems
Email support@cairn-notes.com. Include which Logseq construct came across wrong, and a small example if you can. We read everything.