Thinking States
An agent's status line: the current step shimmers, then rises away as the next one arrives.
- Category: Transitions
- Licence: Charcoal UI Pro
- Built with: React 19, Tailwind CSS 4, Motion
- Dependencies: motion, clsx, tailwind-merge
Examples
Thinking States
API reference
| Prop | Type | Default | Description |
|---|---|---|---|
steps | string[] | — | |
interval | number | 2.2 | Seconds per step. |
index | number | — | Control the step yourself. |
loop | boolean | true |
About Thinking States
Thinking States is an agent's status line: the current step shimmers, then rises away as the next arrives.
Thinking States is an UI transition for React 19, styled with Tailwind CSS 4 and animated with Motion. It drops into Next.js, Vite, Remix or Astro through the shadcn CLI as a single file you own.
When to use Thinking States
- AI agents and assistants.
- Background job progress.
- Build and deploy steps.
- Search progress.
How it works
Steps crossfade vertically and the current one uses a shimmer gradient. Changes are announced.
The change it animates is also announced to assistive tech, and the motion is short enough never to block what the user wants to do. With prefers-reduced-motion on, it holds still or settles instantly instead of animating.
Customizing Thinking States
Every prop is documented in the API reference above with its type and default. The ones you will reach for first are listed here.
The Customize panel next to the preview lets you change interval, loop, size live, and the Code tab updates to match, so the snippet you copy already carries your values.
- steps (string[]).
- interval (number, default 2.2): Seconds per step.
- index (number): Control the step yourself.
- loop (boolean, default true).
Goes well with Thinking States
These transitions pieces share its tokens and motion language, so they sit together on one page without clashing.
- Card Resize: A React box that springs to the size of its content
- Notification Badge: A React notification badge that pops in and slides new counts in
- Page Transition: React page transitions: native-feeling push and pop between views
- Error Shake: A React error shake: a decaying head shake for wrong input
- Arrow Link: A React arrow link whose chevron grows a stem on hover, CSS only
Install Thinking States
npx shadcn@latest add @charcoal/thinking-statesThinking States is part of Charcoal UI Pro. One seat unlocks all 300 Pro components, 107 landing-page sections and 100 full templates, for every personal and client project, with updates included. Add the @charcoal registry once with your access token, then the command above copies thinking-states.tsx into components/ui and installs clsx, tailwind-merge.
You own the file you install: no runtime package, no lock-in, nothing to keep in sync. Change it however your product needs.
See Charcoal UI Pro pricing · Pro setup
Frequently asked questions
- How do I show an AI agent's progress?
- Pass the current step label, such as Searching, Reading or Writing. Thinking States shimmers the current step and lifts it away when the next one arrives, and each change is announced.
- How do I install Thinking States?
- Thinking States is part of Charcoal UI Pro. Add the @charcoal registry to components.json with your access token, then run npx shadcn@latest add @charcoal/thinking-states. It copies thinking-states.tsx into components/ui and installs motion, clsx, tailwind-merge.
- Is Thinking States free for commercial use?
- Thinking States needs a Charcoal UI Pro seat, monthly or a one-time lifetime price. A seat covers unlimited personal and commercial projects, including client work, and every future Pro component.
- Does Thinking States work with Next.js?
- Yes. It is a client component for React 19 styled with Tailwind CSS 4, so it works in the Next.js App Router and Pages Router, Vite, Remix and Astro.
- Can an AI coding agent install Thinking States?
- Yes. Connect the Charcoal UI MCP server (https://charcoalui.com/mcp) to Claude Code, Cursor or any MCP client; get_component returns Thinking States's props and get_install_command returns the exact command.