You started a project on one machine. Now you need another. CRAFT’s device-switch handoff moves both your files and your AI’s memory — no re-explaining, no manual copying, no starting over.
Last week we showed how CRAFT’s git integration automatically backs up every file change. This week: that same git layer is what makes device switching possible.
The Device Problem
You’re deep into a multi-session project on your desktop. You’ve built up context across five sessions — decisions documented, files committed, priorities tracked, lessons learned captured. Then you need to work from your laptop. Or you’re traveling. Or your desktop needs a restart and you want to keep going on another machine.
Without a structured approach, switching devices means manually copying files, re-explaining your project to a fresh AI session, hoping you grabbed everything, and manually reconstructing the context that took five sessions to build. The files might transfer, but the intelligence around them doesn’t.
Handoffs (Week 2) give your AI persistent memory. Git (Week 3) gives your files portable backup. Device switching is what happens when you combine them — your project becomes truly location-independent.
How the Device-Switch Handoff Works
CRAFT’s device-switch protocol is a 4-phase sequence designed to get you working on a new machine with zero re-explanation. It’s handled by a single recipe: RCP-CWK-002a — Device-Switch Handoff (v1.02a).
Before anything transfers, CRAFT verifies the state of your project files on the current device. It checks what’s committed, what’s pending, and what might be at risk. If there are uncommitted changes, it commits them first — nothing gets left behind.
On the new device, CRAFT pulls your latest committed state from the remote repository. Every file, every change, every commit message that documents what happened and why. Your complete project arrives intact.
After the pull completes, CRAFT verifies that everything arrived correctly. File counts, modification timestamps, and git status are all checked against expectations. A partial pull is worse than no pull — you might not realize something is missing until three sessions later.
With files verified, CRAFT runs its standard session initialization — reading the latest handoff, restoring persona state, loading lessons learned, checking project configuration. The AI on your new device now has both the files and the intelligence.
What Makes This Resilient
The protocol includes fallback paths for when things go wrong. Network failures don’t break the process — if git pull fails because wifi is spotty, CRAFT falls back to working with whatever local state is available and flags what couldn’t be synced. You can continue working and re-sync later.
This matters because real-world device switching rarely happens under ideal conditions. You’re switching because you need to leave, because your desktop locked up, because you’re about to board a flight. The protocol handles the messy cases, not just the clean ones.
There are no USB drives involved. No copy-paste between folders. No cloud sync services to configure. If it’s committed to git, it’s available on any device. If it’s captured in a handoff, your AI already knows about it.
The Evidence
CRAFT’s device-switch protocol was tested during the framework’s own development when the project moved from desktop to laptop (Session H009). The full 4-phase sequence executed successfully — files transferred, handoff restored, and the project continued without re-explanation.
The protocol builds on the same infrastructure that’s been proven across 71 consecutive session handoffs with zero data loss. The git layer that enables device portability is the same layer that protected every file change across those 71 sessions. Device switching doesn’t introduce new risk — it leverages systems that are already battle-tested.
Why This Matters
Combined with session handoffs and automatic git backup, device switching completes the foundation. Your AI has memory (Week 2). Your files are safe (Week 3). Your project goes where you go (this week). That’s a working environment that doesn’t depend on where you happen to be sitting.
This is Week 4 of our 8-week capability spotlight. Each week we go deep on one part of CRAFT — how it works, what problems it solves, and how to use it. Follow along as we build the case for structured AI working environments.
