Claude Code just added a new slash command, /ultraplan, which moves the planning phase of your coding workflow out of the terminal and into the browser.
The problem with planning in the CLI is that you’re stuck in a linear, terminal-bound loop.
You run/plan, read the wall of text, approve or reject the whole thing, and move on.
There’s no way to comment out a specific section, paraphrase something you disagree with, or mark something you want to change without rewriting the entire prompt.
/ultraplan fixes it.
You run commands from your local CLI, and Claude creates a remote execution plan on Anthropic’s cloud infrastructure.
While this is happening your terminal remains free.
When the plan is ready, you open it in a browser. claude.ai And interact with it like a document: inline comments on specific passages, emoji reactions per section, and outline sidebars for jumping around.
You can ask Claude to edit, iterate and redraft until the plan looks right.
After approval, you choose: execute the plan in the same cloud session and get PR, or teleport it back to your terminal with access to the full local environment.
What makes it interesting is where the planning takes place and where the execution takes place.
Planning is mostly reading and reasoning, so the cloud is fine.
Execution often requires your local environment. /Ultraplan respects this distinction.
Who is it for?: Engineers and teams already using Claude Code who need more control at the planning stage before code is written.
Especially useful for large refactors, migrations, or anything where blindly approving a project feels risky.
Considerable: Requires Claude Code v2.1.91 or later, Claude Code on the web enabled, and a GitHub repo. Not available on Bedrock, Vertex, or Foundry.