Your work, across AI agents

Centillex Desk

Pack. Move. Work.

Take your work to the next agent, anywhere.

Pack your project workspace, supported settings, and sessions. Move across subscription-based AI vendors and keep working, on your machine or your cloud server.

# Save your work after stopping Claude
$ desk pack --from=claude

# Continue with another agent
$ desk move --to=codex

Workspace. Settings. Sessions.
A saved pack, ready for the next handoff.

Linux · macOS · WSL   /   Native agents on local machines and your cloud VMs

Why Desk?

Your AI strategy should move with your business.

Organizations choose AI vendors for different reasons: capability, subscription budgets, team needs, and internal policy. Those choices change. The work your people have already done needs to carry forward.

Centillex Desk makes that work portable. Teams can move saved progress between the agents their organization chooses, with project files and conversation context available at the next handoff.

Put vendor choice into practice

Adopt a new AI vendor or move to an organization-approved tool without manually rebuilding the project context for every handoff.

Use the subscriptions you have

When one subscription reaches its usage limit, pack your progress and continue with another available, authorized agent. Choose the tool that fits the next task.

Keep work portable

Carry your workspace, supported settings, and session history as a pack you control. Move to another machine as your team’s infrastructure needs change.

Enable your AI-SDLC

From a developer’s desk to your AI Software Factory.

Developers begin with familiar agents and subscriptions. Your organization brings the infrastructure, engineering standards, and delivery process.

Centillex Desk connects those environments. Pack a workspace with its supported settings and session history, then move the work into the agents running on your organization’s machines or cloud servers.

Bring the developer’s work and context into the next stage of your AI-assisted software development lifecycle (AI-SDLC). Desk provides the handoff; your own systems handle orchestration, testing, review, and delivery.

Bring your first project
From one agent to the next

Pack. Move. Work.

A practical handoff for the work behind your AI strategy.

  1. 01

    Pack.

    Stop the source agent and save the project workspace, supported native settings, and sessions in a portable archive.

  2. 02

    Move.

    Choose the next agent. Desk prepares the target workspace and a handoff with your saved context.

  3. 03

    Work.

    Continue in the target agent, then return to its saved session with its own resume command. Pack again when you are ready for another move.

Cross-agent moves start a new conversation with handoff context. Review the target’s settings and authenticate as needed. See supported agents and move directions

Anywhere your team works

Local machines. Your cloud.

Choose the agent and the machine that suit the work.

On your machine

Pack and move between installed agents on Linux, macOS, or WSL. Move with your saved project context. Local Bob handoffs keep your existing project directory; other targets use a restored workspace.

Start a local handoff

On your cloud server

Run Desk and your chosen agent on a cloud VM. Transfer your saved pack to that machine’s Desk store, then move it by pack ID. You choose and manage the host.

Use the same agent workflow on infrastructure your organization controls.

Learn about portable packs
Start with your next handoff

Install Centillex Desk

Linux and macOS on x64 and ARM64. Windows through WSL.

INSTALL WITH NPM
npm i -g @centillex/desk
desk --version

Use Node.js 20 or newer with npm. On Windows, run the installation inside WSL with Linux Node.js and npm. Keep optional dependencies and install scripts enabled.

npm installs a native Go binary with SQLite embedded. Desk runs without Node.js or a separate SQLite installation. Install and configure the agents you want to use separately.

Follow the getting started guide

Keep moving

Report a bug

Share a reproducible issue or request a feature in our public issue tracker.

Open the issue tracker

Your organization’s next step

Talk with us about moving developer work into your own AI Software Factory.

Talk to Centillex