Archive any session. Restore it when you need it.
Keep encrypted copies of ended Claude and Codex sessions, or a snapshot of a running one, in your S3 storage under a dedicated archive key. Archived sessions stay searchable. Free their local history once the copy is verified, and restore it later.
Claude and Codex only, off by default, and archives cannot be recovered without their key. Archives are not compressed. Moving a workspace archive to another Host needs every other session in that container to have ended. Verified against a disposable MinIO bucket; a run against production S3 storage is still pending.
How it works
Create a dedicated archive key
Settings → Session archive creates an AES-256 key separate from the Freeze and Overnight key and shows it once with a check code. Archiving stays off until you confirm its backup. Rotated keys keep older archives readable.
Choose what is archived
Ended sessions keep their native conversations, an allowlisted part of the session home and redacted Task Graph metadata. Opt into the workspace scope to include the repository and checkout. Credential files and stores are never archived.
Free space deliberately
After a verified archive, Free space downloads and authenticates it again and removes only local transcript files identical to their archived copies. A session whose history was freed is restored before it resumes.
Restore or automate
Restore writes only missing files and keeps newer local ones. Optional rules archive old ended sessions or the oldest ones on a Host that runs short of disk, and an optional retention term deletes old archives by their recorded identity.
Try the interface
Explore the sample projects, accounts, files and access requests. Live sessions and file transfers are available in your own console.
Put your server to work.
Give your agents persistent remote workspaces. Start with one repository and server, then add projects, developers and resources from the same console.