Announcing Proliferate
Proliferate

Local & cloud

Where local and worktree workspaces execute, with a preview of Cloud workspaces coming soon.

Proliferate currently runs workspaces on your machine, either in an existing checkout or an isolated git worktree.

TargetRuns onSharedIsolated
Local checkoutYour machine, in the repository's existing checkoutYour installed toolchains, dotfiles, and running processesNothing beyond git; agent edits land directly in the primary checkout
New worktreeYour machine, in a separate git worktree for the same repositoryThe same machine, disk, and installed toolchains as the local checkoutIts own directory and branch, so agent work normally stays out of the primary checkout

Terminals, file browsing, browser previews, and agent commands all run on your machine for both targets. Worktrees isolate branch and workflow state, but are not a security boundary.

Cloud workspaces are coming soon

Cloud workspaces will run repository work remotely, away from your laptop. They are not currently selectable in the product, and setup, authentication, wake/resume, limits, and troubleshooting instructions will be published when the feature becomes available.

Info:

Documentation about deploying the Proliferate server on AWS, Azure, GCP, or Kubernetes describes where a self-hosted control plane runs. It does not mean that product Cloud workspaces are currently available.

Local and worktree disk

A worktree is created with its workspace and shares your machine's installed toolchains and dependencies. Archiving a local checkout or worktree hides it from the active UI; it does not prune files.

Settings -> Pruning inventories Proliferate-managed worktrees. An operator can opt a runtime into retention of old, clean worktrees, or explicitly purge a managed workspace and its runtime history after a safety check. The linked local checkout itself is not a managed prune target. See Lifecycle & storage.

On this page