Overview
Open Input

Chat-style prompt interface for submitting queries to AI

inputpromptchat
Prompt Actions
Props
PropTypeDefaultDescription
placeholderstringundefinedPlaceholder text for the input field
onSubmit(prompt: string) => voidrequiredCallback fired when the user submits a prompt
isLoadingbooleanfalseWhether the input is in a loading/disabled state
suggestionsstring[]undefinedOptional autocomplete suggestion strings
multiModalbooleanfalseEnable file/image attachment support
acceptedFileTypesstring[]undefinedMIME types accepted for file uploads (requires multiModal)
maxLengthnumberundefinedMaximum character length for the prompt input
Docs

Open Input is the foundational interaction point between users and AI systems. It provides a natural language interface — typically a text field — where users express their intent in their own words. While this feels immediately intuitive, it introduces a classic cold-start challenge: users often don't know what to type after their first message. Effective open input design bridges this gap by combining the free-form prompt area with supportive affordances like scope indicators, contextual hints, and graceful error handling.

Variants

  • Chat box A persistent input anchored at the bottom of a conversation thread. Well suited for iterative, back-and-forth dialogue and follow-up questions.
  • Inline composer A prompt that activates at the cursor position within an editor. Ideal for targeted edits such as rewriting a paragraph or generating content at a specific location.
  • Command-style prompt A single-line field paired with structured flags or controls. Favored by power users who want precision and repeatability.
  • Side panel composer A prompt housed in a dedicated panel that can reference files, tools, and settings. Best for complex, multi-source tasks that require extended context.

Use Cases

  • Conversational AI assistants and chatbots
  • Document and content generation tools
  • Code completion and programming assistants
  • Creative tools like image or music generators
  • Search interfaces with natural language queries

Best Practices

  • Define a clear default scope and make scope changes a single-step action so users always know what the AI is operating on.
  • Respond to edge cases with constructive guidance rather than generic error messages or silent failures.
  • Don't assume users are skilled prompt engineers — provide wayfinding aids like suggestions, templates, and inline hints.
  • Keep advanced controls accessible without cluttering the default view; progressive disclosure works well here.
  • Maintain user agency at every stage — always surface options for parameters, model selection, and output formatting.

Search patterns

Search for a pattern, category, or tag