Claude Code simply added a brand new slash command,/ultraplan, that strikes the making plans section of your coding workflow out of the terminal and into the browser.
The issue with making plans in a CLI is that you are caught in a linear, terminal-bound loop.
You run /plan, learn a wall of textual content, approve or reject the entire thing, and transfer on.
There is no method to touch upon a particular segment, annotate what you consider, or flag what you wish to have modified with out rewriting all of the steered.
/ultraplan fixes this.
You run the command out of your native CLI, and Claude generates the implementation plan remotely on Anthropic’s cloud infrastructure.
Your terminal remains unfastened whilst that is going down.
When the plan is in a position, you open it in a browser on claude.ai and have interaction with it like a record: inline feedback on explicit passages, emoji reactions in keeping with segment, and an overview sidebar to leap round.
You’ll be able to ask Claude to revise, iterate, and re-draft till the plan seems proper.
As soon as licensed, you select: execute the plan in the similar cloud consultation and get a PR, or teleport it again in your terminal with complete native setting get entry to.
What makes it fascinating is the break up between the place making plans occurs and the place execution occurs.
Making plans is most commonly studying and reasoning, so the cloud is ok.
Execution steadily wishes your native setting. /ultraplan respects that difference.
Who it is for: engineers and groups already the use of Claude Code who want extra regulate over the making plans section earlier than code will get written.
Specifically helpful for better refactors, migrations, or the rest the place approving a plan blindly feels dangerous.
Value noting: calls for Claude Code v2.1.91 or later, Claude Code at the Internet enabled, and a GitHub repo. Now not to be had on Bedrock, Vertex, or Foundry.
