Shuriken vs Conductor (2026)

We make Shuriken. We tried to be fair; corrections welcome at office@shuriken.sh.

Quick answer#

Pick Conductor if you work on a Mac and want managed cloud workspaces, multiplayer and a shipped iOS app in one paid plan. Pick Shuriken if you also work on Windows, Linux or your own servers, or if you run several Claude or ChatGPT accounts and want your chats to move between them by themselves when one runs out.

Both run coding agents in parallel, each in its own isolated copy of your repo. The difference is where they run and what happens when you hit a limit.

At a glance#

Conductor facts are from its own site, read on 2026-10-22. Each cell links to the page it came from.

What Conductor does well#

It's a focused Mac app. Conductor's pitch fits on one line: run parallel Claude Code, Codex and Cursor agents in isolated workspaces on your Mac, see what each one is doing, then review and merge. If that's the whole job, the product stays out of your way.

Cloud workspaces are built in. On Pro and up, an agent can run in a Conductor-managed cloud workspace instead of on your laptop. The pricing page says there is no extra fee for that today, and that usage-based pricing for cloud compute is planned. If you don't own a spare machine, this is the easiest way to keep work going while your laptop is shut.

Multiplayer is shipping. Pro includes live collaboration for up to five Pro users, and Teams removes the cap and adds an admin portal and central billing. Shuriken's shared projects are an opt-in beta in 5.1, and a member's chats run on the owner's computer.

There's an iOS app. Conductor 0.90.0 (2 October 2026) brought an iOS app: chat with agents and merge pull requests from your phone, working with cloud workspaces. Ours is a web app until the native apps land.

New models show up fast. The changelog reads like a model release calendar: Claude Opus 5.5 on 22 September, GPT-6.1 Sol on 29 September, plus Sign in with ChatGPT the same day, so Codex can run on your ChatGPT plan without an API key.

Where Shuriken is different#

Every desktop, plus your own servers#

Shuriken runs on macOS, Windows and Linux, and it installs headless on a server: Settings › Devices › Add device › A server. The installer sets up the agent CLIs and GitHub sign-in, the server runs as a service, and it only updates itself to signed releases, waiting until no agent is mid-turn.

That is our answer to "keep working while my laptop is closed": a machine you already own, not a managed cloud. Conductor's answer is its cloud. Both are reasonable; they cost and trust differently. And if the machine is a MacBook, there is a third answer with a silly name: Cocaine mode. Close the lid, the agents keep working.

Several accounts per provider, switched for you#

This is the reason Shuriken exists. You add every Claude, ChatGPT or Gemini sign-in you own as a source, plus API keys if you like. When the account in use reaches your switch point (99% of what the app shows, adjustable from 50% to 100%) or the provider reports a limit, your chats move to the account with the most room. A "Switched to …" card in the chat says who took over and why.

A Claude Code chat moves on its very next request, subagents included. Other agents move when their current turn ends. API keys can be a last resort, if you turn that on (it's off by default): they take chats only once every subscription is out, and a spend cap per fallback run ($25 by default, across all the keys that run uses) stops it there. More in source switching.

Only add accounts you own and are entitled to use. Providers' terms may not allow sharing a subscription sign-in.

Session limits#

Claude's 5-hour window starts the first time you use an account, not on a fixed schedule. So when an account's window isn't running, Shuriken sends it one tiny request in the background, once per window, and its window starts. Your limits then come back early, and your chats stay where they are. See session limits.

MCP sign-ins that follow you#

Global MCPs are one list of MCP servers for every project and account. You sign in to Linear or Notion once; the sign-in belongs to the server, not to your AI account, so switching accounts never asks again. There are 57 ready-made connectors, and Claude Code and Codex chats are both handed the list when they start.

Voice built in#

Your speech is turned into text on your own computer by NVIDIA Parakeet by default, and it works offline. Prefer a cloud model? Add your own key for ElevenLabs, AssemblyAI or another live provider. You see a live transcript while you talk, and you can paste text mid-sentence: it lands where you were. See voice.

Where your agents run#

Agents, code and chat history stay on your computers and servers. Shuriken's cloud signs you in, links your devices and relays end-to-end encrypted messages. It runs no agents and stores no code.

Pricing#

Conductor prices in US dollars, Shuriken in euros, VAT included for EU consumers.

Neither app bills you for model usage. Your agents use your own plans and keys on both. Full details on our pricing page.

Using them together#

You can. Each app keeps its own isolated copies of your repo, so a Conductor workspace and a Shuriken chat won't collide on one project. The honest version: the two overlap a lot, and you will probably end up living in one. If your team is on Macs and loves Conductor's cloud, a teammate on Windows or Linux can still work the same repo from Shuriken on Windows.

For a three-way look that includes Superset, see Superset vs Conductor vs Shuriken. If accounts are your main pain, start with source switching, then download Shuriken.