The open request is given structure — fields, slots, controls, quality feedback — so the actor composes a complete request without having to already know what completeness looks like.
Forms
Recurring forms along an abstraction ladder: example prompts and fill-in-the-blank passages, structured prompts with labelled fields, paginated composition, configured controls (tone, persona, parameters) replacing free text, and pre-submission quality feedback or enhancement.
Resources & references
- aiuxpatterns.com — Structured Prompt, Paginated Prompt, Cloze Passage, Configurable Controls, Prompt Quality Feedback
- Shape of AI — Templates, Madlibs, Prompt enhancer
- AI UX Playground — Prompt Templates, Tone Sliders, Persona Selector
To-do
Prompt hosts stubs of this territory (structured prompt, quality feedback, templates); migrate them here as this matures. The novice↔expert fit that aiuxpatterns.com tracks per pattern is a contextual variable worth keeping.
Consequences
- the request's expected shape is externalised — labelled fields, fillable slots, controls — so completeness is inspectable before anything runs
- structure narrows expression: what the scaffold doesn't ask for tends not to get said, and the actor's stranger intents stop fitting the form
- a scaffold encodes somebody's yesterday's idea of a good request, and keeps proposing it until the encoding is revisited
Related patterns
Instantiates
Enacts
- Learnability — occupying the scaffold teaches what a good request contains
Complements
- Suggestion — starters and example prompts seed the scaffold's slots
Related
- Wizard — paginated scaffolds are the wizard applied to prompting
- Form — at full structure the scaffold simply is a form whose output is a request
- Validation — prompt-quality feedback is validation applied before the request runs
Enabled by
- Bounded choice — each slot of the scaffold is an act of bounded choice over one part of the request