Work together.
Without commits.
Continuous sync for code, models, and textures. Corotation publishes your saves in the background and keeps automatic checkpoints on your own server.
Open source. Self-hosted. Written in Rust.
Large assets, without a fresh copy for every edit.
Corotation reuses chunks across files and versions. Only chunks the server doesn’t already have need to be uploaded and stored.
See the measurementsIllustrative edit. Re-encoding can change most of a file. How storage differs ↗
Let your agent handle conflicts.
Codex or Claude can repair rejected edits and resume sync.
Agent setupA way back, without a commit history to maintain.
The server keeps immutable baselines every five minutes by default. Recover files from a checkpoint or use it to catch up after time away. Unchanged chunks stay shared.
Working with checkpointsOne command while you work.
cor work
Run current on your server and connect your project once. After that, just start Corotation and keep working in your usual tools.
Set up your first workspaceA few practical questions.
Do we need Git?
No. Corotation works with ordinary folders and respects
.gitignore even without a Git repository. There are no
commits, branches, or code reviews.
What happens when edits compete?
The first valid package wins. If any file in a competing package
conflicts, the entire package is rejected. Your local work is
preserved and recorded in
conflicts.laplace.json for you to repair. Corotation does
not merge it for you.
Is there a paid service?
No. Corotation is MIT licensed and runs on your infrastructure. You cover your own hardware, storage, and network costs.
Will every edit make a small transfer?
Not always. Re-encoding a compressed model or texture can change many chunks. Modified files also need to be read locally to find the changes.
What should we know before using it?
This is an early release. Chunks, baselines, and recovery archives are not garbage-collected yet. Read the current limits before connecting your project.