Conductor and Claude
Conductor vs Claude Code: what Conductor added, and what Claude Code does now
Conductor was a Mac app that ran Claude Code and other agents in parallel, each in its own git worktree, with a diff viewer, a checks tab and cloud workspaces. Claude Code is Anthropic's agent itself. Its --worktree flag and desktop app now run sessions side by side, covering much of what Conductor added.
Ipseeta PriyadarshiniPublished
How do Conductor and Claude Code compare?
| Conductor | Claude Code | |
|---|---|---|
| What it is | A Mac app that runs other coding agents in parallel | Anthropic's coding agent: a CLI, a desktop app, web and mobile |
| Agents | Claude Code, Codex, Cursor and OpenCode | Claude models only |
| Parallel work | One workspace per task, each a git worktree on its own branch | claude --worktree in the CLI; parallel sessions in the desktop app, with an optional Worktree mode |
| Platforms | Mac, plus an iOS app | macOS, Windows, Linux in beta, web, iOS and Android |
| In the cloud | Conductor Cloud, shutting down on November 20, 2026 | Cloud sessions on the web or with claude --cloud |
| Sign-in | Your Claude plan, a ChatGPT plan or API keys | A Claude plan, or an API key for the CLI |
| Price | Free, Pro $50 a month, Teams $60 a user (details); not sold since November 4, 2026 | Included with Claude Pro, from $20 a month |
| Status | Bought by Meta; the Mac app gets critical bug fixes until early 2027 | Actively developed |
How did Conductor run Claude Code?
Conductor bundled Claude Code and ran it through Anthropic's Claude Agent SDK. Its providers guide lists four ways to sign in:
- a Claude Pro or Max plan
- Claude Console or API billing
- an existing Claude Code login
- an Anthropic-compatible provider, like Bedrock or OpenRouter
Conductor didn't bill for model usage.
In May 2026 Anthropic announced changes to how subscriptions work with the Agent SDK in third-party tools. On June 15 it delayed them indefinitely, and Conductor wrote that users could keep using their Claude plan with no extra limits.
What did Conductor add on top of Claude Code?
- Workspaces as worktrees. Each task got an isolated copy of the repo on its own branch, with its own terminal and ports, all managed in one window.
- Review before merge. A diff viewer with line comments sent back to the agent, and a Checks tab with CI status, deployments and GitHub review threads.
- Repo scripts. Setup, run and archive scripts in
.conductor/settings.toml. The run script starts your app when you click Run. - Other agents in the same window. Codex, Cursor and OpenCode next to Claude Code.
- Issues to workspaces. Start a workspace from a Linear or GitHub issue.
- Cloud and mobile. Conductor Cloud from July 2026, then an iOS app and remote control for your Mac's workspaces.
What does Claude Code do on its own now?
Most of the parallel workflow is built in:
- The CLI:
claude --worktree <name>starts a session in its own worktree on its own branch, and cleans it up when you're done. It arrived in version 2.1.49. Claude Code worktrees covers the details. - The desktop app: runs several sessions from its sidebar, and gives each its own worktree when you switch Worktree on (docs).
- The cloud: each
claude --cloudcommand starts its own cloud session, sharing your plan's limits (docs).
What Claude Code doesn't do is run Codex, Cursor or OpenCode next to Claude. For that you need a multi-agent desktop app; Conductor alternatives lists them.
Which should I use now?
You can't sign up for Conductor any more, so for new work the choice is Claude Code itself or another tool. If you only use Claude, Claude Code's desktop app or --worktree replaces most of Conductor. If you ran several agents side by side, a multi-agent app like Orca or T3 Code is closer. If you relied on Conductor Cloud, you need a cloud agent before November 20, 2026.
Where Buildful fits
Buildful is closer to Conductor Cloud than to either desktop app: tasks run in cloud sandboxes and come back as pull requests, with your checks run and UI changes tested in a real browser. Smart Mode picks the right open-weight model for each task. On Pro and Team, you can also pick a frontier model like Claude or GPT, with prepaid credit or your own API key. Pro is $10 a month for 1,000 tasks, and Team is $40 for 3,000 shared by the team (pricing). See Buildful vs Conductor and Buildful vs Claude Code.
Frequently asked questions
Did Conductor use my Claude subscription?
Yes, if you signed in with it. Conductor ran Claude Code through Anthropic's Agent SDK and accepted a Claude Pro or Max plan, an API key, or an existing Claude Code login. It didn't charge for model usage.
Can Claude Code run agents in parallel without Conductor?
Yes. Run claude --worktree <name> in several terminals, each with its own name, or open several sessions in the Claude Code desktop app with Worktree on.
What's the difference between Conductor and the Claude Code desktop app?
Both run parallel sessions in worktrees. Conductor also ran Codex, Cursor and OpenCode, and had a checks tab and Conductor Cloud. The Claude Code desktop app runs Claude only, on macOS, Windows and Linux.
Is Conductor better than Claude Code?
They did different jobs. Conductor managed several agents, including Claude Code, in one Mac app. Claude Code is the agent. Conductor is closed to new users, so Claude Code is the one you can start using today.
What happens to my Conductor workspaces?
Cloud workspaces become read-only on November 13 and are deleted from November 20, 2026, so export them or transfer them to your Mac first. Moving off Conductor Cloud has the steps. Local workspaces are git branches in your repo, so push the ones you still need.
About the author

Ipseeta Priyadarshini
Senior Software Engineer, Bug0
Ipseeta is a senior software engineer at Bug0, with more than ten years of full-stack work in Node.js, Next.js, TypeScript and Playwright, and Java and Spring before that. She has built software at Wipro, DXC Technology, Majesco and Hashnode, and studied Information Technology at IIIT Bhubaneswar. She has wired AI-powered APIs into full-stack apps, and writes about Playwright, end-to-end testing and full-stack engineering.
Try it on your own repo
Describe a change. Buildful builds it, tests UI changes in a real browser, and opens a pull request. Your first 3 tasks are free, then 1,000 tasks for $10 a month. See pricing.
Start building