Inputs & Forms: AI Skill for UX Design
Inputs & Forms is an AI skill that builds forms people finish: minimal fields, one column, labels outside inputs, and kind inline validation. Extras like multi-step splitting stay gated behind trade-offs.
npx skills add Uxcel-Lab/product-skills --skill ux-inputs-and-forms
What this skill does
Helps an AI assistant build forms that people finish instead of forms that over-ask, mislabel, and validate late.
- Cuts every field the task doesn't require and marks required and optional ones.
- Lays fields out in one column with logical groups.
- Keeps labels above inputs and always visible, never placeholder-only.
- Picks the input type from the data: radio buttons under five options, masks where format matters.
- Validates on blur with specific, polite messages at the erroneous field.
- Labels the submit button with the action, not "Submit".
- Shows loading, success, and failure states after submission.
- Gates multi-step splitting, floating labels, and helper text as choices with trade-offs.
When to use it
Use it when you need an AI assistant to:
- design a signup, checkout, or survey form;
- cut a form down to the fields that matter;
- pick the right input type per field;
- design validation and error states;
- write clearer labels, helper text, and error messages;
- adapt a form for mobile;
- review an existing form for completion blockers.
Decisions this skill helps you make
| Decision | Options |
|---|---|
| Multi-step splitting | Single page until grouping stops working, steps from three stages |
| Floating labels | Static labels above inputs, floating as a space-saving variant |
| Helper text | Only where users predictably hesitate, one line maximum |
| Auto-suggest | For any option list users would have to scroll |
| Smart defaults | Prefill facts, never decisions |
| Full-name field | One name field unless the data model needs two |
| Error summary | Inline always, a top summary added for long forms |
| Verification | Risk-based and accessible, never block paste |
How the skill works
1
Justify every field
The skill asks what each field buys and cuts anything the task doesn't require.
2
Apply the core
It lays out one column, keeps labels visible, picks input types from the data, and validates on blur.
3
Surface trade-offs
It presents multi-step splitting, floating labels, helper text, and smart defaults as choices with trade-offs.
