Labsco
MCP SERVER

Rakit UI AI

by madebyaris

Put three HTML versions of the same component side by side with the framework, style preference and viewing mode set on the call, and record why one of them won.

Design Systems & UI Component LibrariesVerified
Summary
The reasons travel with the candidates, so the choice leaves a record of what it was chosen against.

purpose_include, purpose_exclude and style_preference are part of the same call as the three markups, which is the part of a design decision that normally evaporates once the chat scrolls. The tool stops at selection: it compares and returns a choice, and nothing here writes the winning component into your codebase.

What it is

A single-tool comparison surface for UI components: three candidate markups go in with their names, and the tool adapts the layout it shows them in.

What you get
  • A comparison of 3 UI component designs, passed as design_html_1, design_html_2 and design_html_3 with a name for each.
  • A viewing mode chosen from the component's complexity, or forced through view_mode.
  • Framework context on the call — the tool names Tailwind, Bootstrap and Bulma among the CSS frameworks it supports — so candidates are judged as the project would render them.
  • The intent captured alongside the markup: purpose_include, purpose_exclude, purpose_notes, style_preference and component_type.
  • Design generation from a natural-language prompt through the MiniMax-M2.1 API, and reference lookup through auto_search_reference.
Requirements

Nothing to supply — no account, no key. The markup for the candidate designs, unless you are having them generated from a prompt.

Setup effort

One command — npx -y rakitui-ai