The Claude Code Desktop Redesign Is Not What You Think It Is

There's no code editor.

That's the first thing you notice when you open the redesigned Claude Code desktop app Anthropic shipped on April 14. No autocomplete pane, no go-to-definition, no multi-cursor. What you get instead is a sidebar, a chat panel, and a place to watch things happen.

One developer on X put it perfectly: "The most revealing detail about the redesign: there's no code editor anywhere in the main interface."

He meant it as a criticism. It's actually the whole point.

What Anthropic actually built

Anthropic's own framing is direct: "The new app is built for how agentic coding actually feels now — many things in flight, and you in the orchestrator seat."

That's not a tagline. It's a description of the mental model shift. You're not writing code with AI assistance. You're running multiple AI sessions in parallel, watching their progress, steering when something drifts, and pulling finished work back when it's ready.

The headline feature is the multi-session sidebar. Left panel shows every active and recent session. You kick off a refactor in one repo, a bug fix in a second, a documentation pass in a third, a new feature in a fourth — all from one window, all running at the same time. Each session gets its own context window and its own permission scope.

One developer who spent two full days inside the new app described the shift like this: running four parallel sessions changes the workflow more than all the other new features combined. You kick off a refactor in session 1, then start a bug fix in session 2 while session 1 is still thinking. By the time the bug fix is ready for review, session 1 has finished and you're already on session 3. The difference between a junior dev doing one task at a time and a tech lead with four things in flight — that's the gap this app is trying to close.

What else shipped

The integrated terminal and file editor are less flashy but more useful day-to-day. Before this update: Claude generates code → you switch to your editor → you switch to your terminal → you copy results back. That context-switching was the friction nobody talked about but everybody felt and that the issue. Now the terminal runs inside the app. The file editor opens files, lets you make spot edits, saves changes. The diff viewer got rebuilt from scratch for performance on large changesets.

There's also side chat — triggered by Command + ; — which lets you ask a question off a running session without feeding extra context back into the main thread. This solves a specific annoying problem: you're seeing a long session run and want to ask something quick, but typing into the main thread interrupts the agent's context. Side chat is its own clean window that knows about the main session but doesn't redirect it.

Three view modes — Verbose, Normal, Summary — control how much of Claude's tool-call activity you actually see. Summary is the one most people will want for fire-and-forget sessions. Switch to Verbose when something looks wrong and you need to see every argument.

Sessions now auto-archive when the associated pull request gets merged or closed. No more 200 abandoned sessions clogging the sidebar.

SSH support now extends to Mac alongside Linux. If you run agents on always-on remote machines, you no longer need custom scripting to get the graphical interface working.

Routines shipped the same day

Same announcement cycle, April 14: Routines — a scheduled automation system for Claude Code. You define a prompt, a repo, and some connectors, and the routine runs on Anthropic's cloud infrastructure on a schedule or triggered by a GitHub event. Your laptop doesn't need to be on.

Pro users get 5 routine runs per day. Max gets 15. Team and Enterprise get 25. It's in research preview, so expect the limits and behavior to change.

The way one builder framed it: Routines handle the unattended recurring work, the desktop handles the interactive work. Together they cover most of what a small dev team currently does manually.

Who it's actually for

If your workflow is "I write code and AI helps me," Cursor is still smoother. The editor experience, autocomplete, inline suggestions — Cursor owns that pattern. Most developers who've tried the new Claude Code desktop report keeping both: Cursor for active coding, Claude Code for directing agents.

If your workflow is already "AI does the work and I review it," this app was built exactly for you. Four parallel sessions on a real codebase is a fundamentally different way of working than anything Cursor or Copilot currently supports.

The token cost is real though. Four sessions running simultaneously means four context windows burning. If you're on Pro at $20/month, heavy parallel use will hit your shared limit faster than you'd expect. The people this app is most useful for are probably the same people who should be on Max.

Available now for Pro, Max, Team, and Enterprise. Download at claude.ai or update and restart if you already have the app. You need desktop v1.2581.0 or later. Linux support is coming in the next few weeks.