What's new in Restora

restora export-comments <page-url> pulls whole comment threads out to Markdown or CSV — including the inline ones attached to individual blocks, which a naive export misses entirely. Free, and it runs on your machine. How it works →

Every backup now tells you precisely what it captured: databases, rows, pages, views, relation links, files and comments. Anything missing is named, not rounded away — a skipped attachment shows its filename, its size and the reason it was skipped. There are no percentages anywhere, on purpose: "99.8% backed up" hides the one file you needed.

The Restore Simulator moved to the front of the dashboard. It shows how many pages a backup would rebuild and how many relations it would reconnect, plus a realistic time range based on real measured restores — never a made-up single number.

Pro now backs up files up to 150MB each, Free up to 20MB. It applies to your existing schedule automatically — there is nothing to turn on, and no re-connect needed.

When an attachment doesn't make it into a backup you get the filename, the size, and the page it lives on — and files that were too large are reported separately from files whose download failed, because only one of those is a limit. The second kind usually succeeds on the next run.

Network failures used to collapse into one unhelpful line. The CLI now tells you which part of the connection actually failed, and supports proxy configuration for locked-down company networks.

A downloadable record of what was rebuilt, what was checked, and anything that didn't come back — written as verified facts rather than a green tick. Keep it as proof the recovery happened.

Restored boards used to arrive with every status piled into a single column. They now rebuild with their original groupings intact.

A focused round of performance work on the biggest workspaces: backups, restores and the database picker all got substantially quicker, and workspaces with thousands of databases stopped struggling.

Backups can now be cancelled mid-run — and a backup you deliberately stopped is no longer reported back to you as one that failed.

Large backups running in your browser now pick up where they left off after a dropped connection, instead of starting over. Restores that lose their tab reconnect on their own, and finish server-side either way.

restora backup --incremental re-reads only what changed since your last backup, so daily runs on a big workspace finish in a fraction of the time.

Rebuilt around a single question you can answer at a glance: if something went wrong right now, could you get it back? Status, timeline, what's protected and what isn't, in one place — and it never shows green unless that's actually true.

Pro runs your backups on a schedule from our servers, straight into your own Google Drive or S3 bucket. Same store-nothing rule as always — the file never stays with us. You get an email if a run fails.

Every backup becomes a restore point. Scroll the timeline, pick any date, and restore the whole workspace or a single database from exactly that moment.

Restora watches your live workspace against your last backup and alerts you when a database is deleted or emptied — so you find out the day it happens, not the month you go looking for it.

Big restores no longer depend on your browser staying open. They run server-side from your own cloud storage and email you when they're done — close the tab, go to lunch.

Every backup is integrity-checked the moment it lands — every referenced file present and decodable, every in-backup relation resolving — and the result is stamped on its timeline entry. And when you do restore, a one-click rehearsal audit re-reads the restored copy and hands you the receipt.

restora guard takes a snapshot before an agent runs and shows you exactly what it changed afterwards. Free, and it runs entirely on your machine.

Restored content lands under a name you choose instead of a raw timestamp, with a note explaining where it came from and when.

Turn any Restora backup into a real Obsidian vault, or into spreadsheets that open cleanly in Excel and Sheets with relations resolved to readable names. Free, and never paywalled — leaving should be as easy as arriving.

Standalone pages back up and restore with their hierarchy rebuilt between them, and a single explorer view lets you pick everything you want protected in one place. You can also search inside a backup.

Point Claude or any MCP-compatible assistant at your local backups and ask questions about them. It can read, and only read — your live Notion workspace is never exposed.

$9.99 unlocks restoring your backups for 30 days, with no subscription and nothing to cancel. Pro is still there if you want ongoing automatic backups and monitoring.

Compares your backup against the live workspace and lists every structural difference it can see: deleted databases, emptied tables, removed properties, retyped fields. Free, read-only, and it never touches your live data.

A free sandbox restore rebuilds sample databases in front of you, relations and rollups reconnecting live, so you can judge whether it actually works before entering a card.

Attachments in file properties and file blocks are backed up and put back on restore, along with the page icons and covers that make a workspace recognisable.

The first working version: back up a Notion workspace and put it back with its relations, rollups and views intact — the part a Notion export leaves out. Store-nothing from day one, and the command-line version has been free and open source since this commit.