Pairing Mode
Mirror an agent’s container into your IDE so you can inspect, run, and edit code locally.
What is Pairing Mode?
Pairing Mode lets you open and work with the agent's container code locally.

When you turn it on, Sculptor:
Checks out the agent’s branch on your machine
Syncs file changes between the container and your local directory
Lets you use your IDE (Cursor, VS Code, JetBrains, etc.) to view or modify the agent’s work
Tip: Pairing Mode is the best way to inspect generated code, run the app locally, make small fixes, or debug what the agent produced.
How Pairing Mode works
When activated, Sculptor:
Ensures your local repo matches the agent’s base commit
Checks out the agent’s branch locally (e.g.
sculptor/refactor-to-tailwind-css)Mirrors file changes live
Restores your previous branch when Pairing Mode is turned off
Your editor will show the exact files the agent sees inside its container.
Starting Pairing Mode
From the toolbar
Open any agent session
Click the Pairing Mode toggle in the top-right corner

From the agent task view
Open the agent from the left sidebar
Go to Task
Click Start Pairing Mode
Status indicators
When Pairing Mode is active:
The toolbar shows Pairing On
Hovering reveals Stop Pairing
A bottom status bar shows:
The synced local path
The container branch currently checked out

Stashing
Pairing Mode includes built-in stashing so you can start pairing without first cleaning up your local Git state. When enabled via the dedicated header button, Sculptor temporarily stashes your current local changes—similar to git stash and switches your workspace to match the agent’s branch exactly.
This lets the agent’s container changes appear cleanly in your local repo without conflicts or manual intervention. When you stop Pairing Mode, your original branch and stashed changes are automatically restored.
The value: you can jump into Pairing Mode instantly, inspect or debug the agent’s output, and return to your in-progress work without losing context.
Summary
Pairing Mode gives you a safe, familiar local workflow while still leveraging Sculptor’s containerized agents. It’s ideal for:
Reviewing large changes
Running code in your own environment
Fixing or tweaking generated results
Debugging behavior before merging
Last updated