Comparison · Last updated 2026-08-08
Logseq moved to a database.
SeqLog stays on your files.
On 13 July 2026, Logseq 2.0 shipped in beta with a local SQLite database as the canonical source of truth. The file-based build was renamed Logseq OG and now receives security and Electron updates only — no new features. If plain Markdown is why you chose Logseq in the first place, this page is about what your options are.
The short version. SeqLog is an Apple-native outliner where Markdown files are the only data store — no database, no index, no import step. It opens a Logseq graph directly. It is macOS-only, has no plugin ecosystem, and is not open source. Logseq has far more features, runs everywhere, and has a real community behind it.
Pick SeqLog if file ownership and native speed are what you are missing. Stay with Logseq if you need plugins, queries, Linux or Windows.
What actually changed in Logseq 2.0
Logseq split into two products. Both are still Logseq, and this is not a criticism of either — it is simply a fork in the road that matters if you picked the app for its file format.
- Logseq (DB version) keeps the name and the roadmap. Notes live in SQLite. It brings typed properties, tags-as-classes, dashboard queries, real-time collaboration sync and a new iOS app. Desktop is still Electron; mobile is Capacitor.
- Logseq OG is the file-based build you have been using. It is maintained — security fixes and Electron upgrades — but new features go to the DB version.
For most people the DB version is an upgrade: it is faster on large graphs and it makes features possible that files made awkward. The people it does not serve are the ones for whom the files were the feature.
Side by side
| SeqLog | Logseq 2.0 (DB) | Logseq OG | |
|---|---|---|---|
| Source of truth | Plain .md files | SQLite database | Plain .md files |
| Runtime | Native SwiftUI + AppKit | Electron | Electron |
| Platforms | macOS 15+ only | macOS, Windows, Linux, iOS, Android, web | macOS, Windows, Linux, mobile |
| Search | ripgrep over the files, no index | Database queries | Built-in index |
| Version control | Git built in — auto-commit, push/pull, diffs, revert | RTC sync service | External Git, by hand |
| Plugins & themes | None | Ecosystem | Ecosystem |
| Query language | No | Yes, stronger than OG | Yes |
| Editable by scripts & agents | Yes — it is just files | Not directly | Yes |
| Open source | No | Yes | Yes |
| Price | Free forever | Free; paid sync | Free |
Why files, concretely
“Local-first” gets said a lot. Here is what having actual files on disk buys you, and each of these stops working the moment the source of truth moves into a database:
- Your agent can work in your vault. Claude Code, Codex, a shell script, a cron job — anything that reads and writes files can read and write your notes. SeqLog watches the folder and picks up external edits. This is the reason that came up most often when Logseq users discussed the change, and it is not a niche one.
- Any sync you already trust. Git, Syncthing, rsync, a NAS, iCloud Drive. Files sync; a live database does not, at least not safely.
- Diffs that mean something. A commit shows the sentence you changed, not an opaque binary blob. SeqLog's Git integration shows those diffs inline and reverts down to a single file.
- An exit that costs nothing. There is no export, because there is nothing to export from. Delete the app and the folder is still a folder of Markdown.
Moving over
There is no import. SeqLog uses the same layout Logseq does — a pages/ folder and
a journals/ folder — and it reads Logseq's YYYY_MM_DD journal filenames
alongside its own YYYY-MM-DD. Point SeqLog at your graph folder and your notes,
wikilinks, tags and daily journals are simply there.
Because nothing is converted, this is also reversible: your Logseq install keeps working on the same folder. Try both on the same vault for a week and keep whichever you reach for.
Two honest caveats. SeqLog reads CommonMark plus wikilinks; Logseq-specific
syntax it does not implement — {{query}}, {{embed}}, logbook blocks,
advanced properties — is left in the file as literal text rather than rendered. And if your graph
is already on the DB version, there are no Markdown files to point at; export from Logseq
first.
Where Logseq is the better answer
A comparison page that concludes “buy mine” for everyone is worth nothing, so plainly:
- You use Windows or Linux, or want the same app on your phone today. SeqLog is macOS-only; an iOS version is in progress.
- You rely on plugins, themes or the query language. SeqLog has none of those, and there is no plugin API planned.
- You want open source. Logseq is; SeqLog is not — it is free, but the source is not published.
- You want a large community writing guides and answering questions. SeqLog is a one-person project with a young user base.
Questions
Can SeqLog open my existing Logseq graph?
Yes — same folder layout, and it understands Logseq's journal filenames. Nothing is converted, imported or copied.
Does SeqLog use a database?
No. No SQLite, no cache, no search index. Search greps the files each time through a Rust ripgrep FFI, so there is nothing to rebuild and nothing to go stale.
Is SeqLog open source?
No. It is free forever on macOS, but the source is not published. If open source is a requirement, this is the wrong app for you and that is a fair reason to stop reading.
Can Claude Code or Codex edit a SeqLog vault?
Yes. It is a folder of Markdown files. SeqLog notices external edits and picks them up.