Has anyone had Replit recover lost code/backups? Agent has cost me 2 months of work 🤯

Bugs
1

I’m honestly at my wits’ end with Replit right now and hoping someone in the community has some advice.

I’ve spent around AU$14,000 with Replit overall building my production app, Advanced Planner Designer, and after repeated problems caused during Replit Agent work, I’m now missing approximately two months of development.

This wasn’t one Agent mistake. Agent caused serious problems with my source/files, I spent days recovering the app and eventually got it deploying successfully again, only for further Agent work to cause more problems and deployment to fail again.

The biggest disaster is Git.

For those two months, Replit showed my project as connected/synced to Git and I was committing my work, so I believed my recent code was protected.

When everything went wrong, I discovered it wasn’t actually synced like I thought it was.

The last usable Git commit was approximately two months old.

So after the newer source was lost, APD effectively regressed back to that old Git version, taking two months of features, fixes and paid development with it.

Major features I had already built are gone, including Photo-to-Colouring and background planner generation, along with loads of work completed after them. I’ve also lost thousands of templates during this whole recovery nightmare.

To make matters worse, Agent also disrupted/deleted the local “.git” state during the recovery problems, so I can’t simply rollback to the recent version I thought I had.

Replit Support has been absolutely shocking.

I first contacted Support 11 days ago!!

I’ve explained the seriousness of the situation and given them everything they need. I’ve been told my case has been “forwarded to the appropriate team”…

And I’m still waiting.

I haven’t heard from that team once.

No meaningful technical update. No indication whether they’re actually investigating. No answer about whether my missing code can be recovered. Nothing.

Considering this involves a production app, repeated Agent-caused problems and approximately two months of paid development disappearing, I honestly expected a hell of a lot better than being left waiting for for over 10 days in counting…

So I’m asking you guys instead :folded_hands:

Does anyone know if Replit keeps additional backups that we can’t access ourselves?

Old workspace backups? Previous versions? Deleted files? Agent history/checkpoints? Internal snapshots? Anything?

Has anyone ever had Replit successfully recover code when Git didn’t have the version you needed?

And does anyone have any other recovery suggestions I should try?

The newer code absolutely existed on Replit before this happened. I just need to find out whether there’s another copy of it somewhere.

If you’ve dealt with anything remotely similar, please tell me what worked for you.

At this point I’m feeling as if I’ll get useful ideas from other users than I ever will for Replit Support ATM. :sob:

— Caris

2

Hi Caris — sorry, that’s a brutal one. A few things you may not have tried yet, roughly in order of how likely they are to work.

Before anything else: stop running Agent on that app and don’t redeploy. Every Agent run writes over whatever is still recoverable. Fork the app now so you have a frozen copy to work from.

  1. gitsafe-backup. Replit keeps an automatic server-side Git backup for every app, separate from your local .git and from GitHub. It syncs when Agent makes changes and when you publish, and it exists specifically for corrupted or deleted .git folders. Agent nuking your local .git doesn’t touch it. In the Shell:

    git remote -v
    git fetch gitsafe-backup main
    git log gitsafe-backup/main

Check the dates in that log — if your Agent work was committing over those two months, the newer commits should be there. If .git is too broken to run those, you can clone the backup directly:

git clone --bare git://gitsafe:5418/backup.git /tmp/gitsafe-restore
mv .git .git_corrupted_backup
mv /tmp/gitsafe-restore/ .git

Keep the corrupted folder, don’t delete it.

  1. Deployment history. You said APD was deploying successfully during that period. Each successful deploy is a snapshot of the source at that moment, living outside your workspace. Check your deployment history for a build from the newer period and roll back to it. Even if that only restores the running build rather than the source tree, a frontend bundle with source maps (.js.map files in dist/ or build/) reconstructs the original source almost exactly. Worth looking in dist/, build/ or .next/ in the workspace too — those artifacts often survive a source regression.

  2. Roll forward, not just back. Agent checkpoints go both directions. If you rolled back during the recovery attempts, the later checkpoints still exist as an alternate branch in the history view and you can move forward into them again — but only until new changes overwrite that branch. Another reason to freeze now.

  3. git fsck. If the old .git is still on disk in any form:

    git fsck --lost-found --unreachable

That surfaces dangling commits and blobs no branch points at. “Deleted” usually means refs removed, not objects actually collected.

On Support: you’re in Australia, so you can file a formal Privacy Act APP 12 access request for all personal data Replit holds on your account. That’s a statutory deadline rather than a support queue, and it tends to get a response. It’s leverage on the silence, not a recovery method — but 11 days with nothing is not acceptable on a production app.

Hope something in there lands. Post back on the gitsafe-backup log output if you get it — that’s the one I’d bet on.