Store Checkpoints with Your Code - Entire

By default, Entire stores checkpoint data in the same Git repository as your code. This keeps the commit, checkpoint, and agent context together while leaving your normal Git history clean.

Choose a checkpoint backend

The checkpoint backend controls how Entire organizes checkpoint data inside Git.

Backend How it stores checkpoints Checkpoint ID
Branch-based (default) Stores all checkpoints on the shared entire/checkpoints/v1 branch 12-character hexadecimal ID, such as a3b2c4d5e6f7
Ref-based Stores each checkpoint in its own Git ref 26-character, time-sortable ID, such as 01KVBJCWYA4YW6J5M9GP655HZN

Branch-based storage

Branch-based storage requires no additional configuration. Entire writes checkpoint data to entire/checkpoints/v1 and pushes that branch alongside your code. To switch an existing repository back to branch-based storage, run:

entire configure --checkpoint-backend branch

Ref-based storage

Ref-based storage lets Entire push and fetch checkpoints independently instead of updating one shared branch. This keeps checkpoint operations fast as agent history grows and lets multiple agents save checkpoints at the same time without competing to update the same branch. For a new repository, run:

entire enable --checkpoint-backend refs

For an existing Entire-enabled repository, run:

entire configure --checkpoint-backend refs

Changing the backend affects new checkpoints. Existing checkpoints remain on entire/checkpoints/v1 and continue to work alongside new ref-based checkpoints. Your workflow and Entire commands stay the same.

Add --local to a configuration command when the backend choice should apply only to your clone. See Where storage settings are saved.