AI Components Overview

A comprehensive suite of 54 modern AI primitives, conversation bubbles, streaming reasoning, tool inspectors, and agent execution interfaces.

The @celestia-project/ui library ships with a comprehensive suite of 54 production-ready AI components built to power generative AI applications, agent workflows, LLM chat interfaces, multi-modal audio/speech interactions, and live sandbox executions.

All AI components are fully styled with Tailwind CSS v4, support light and dark theme OKLCH tokens, and provide keyboard and screen-reader accessibility.


Installation & Import Conventions

You can import AI components from the main package barrel @celestia-project/ui or via deep imports @celestia-project/ui/components/ai/<name>:

tsx
// Package root importimport {  PromptInput,  Reasoning,  Tool,  ModelSelector,  Plan,  Task,} from "@celestia-project/ui"
// Dedicated AI submodule importimport {  PromptInput,  Reasoning,  Tool,} from "@celestia-project/ui/ai"
// Deep component importimport { PromptInput } from "@celestia-project/ui/components/ai/prompt-input"

Live Showcase

Below is a quick interactive preview of key AI components composed together in an agent interaction loop:

The user is requesting an automated PostgreSQL migration for post tagging. Analyzed schema in packages/db/src/schema/tags.ts and computed foreign keys.

Parameters

{  "migration": "0002_add_posts_idx.sql",  "target": "postgresql"}

Result

{  "applied": 3,  "durationMs": 42,  "status": "SUCCESS"}

Component Taxonomy

The 54 AI components are organized into 6 core functional domains:

1. Prompts, Inputs & Voice

ComponentDescription
prompt-inputExpandable AI prompt bar with file dropzone, model selector, tool buttons, and action menu.
chat-inputAuto-resizing textarea specialized for standard chat input prompts.
speech-inputMicrophone voice recording input with real-time waveform visualizer.
suggestionQuick-prompt recommendation pills and action chips.
mic-selectorInput audio device selection dropdown.
voice-selectorVoice profile picker with audio test samples and gender indicators.

2. Chat & Streaming Message Experience

ComponentDescription
messageFull-featured message container with markdown rendering, branching history, and action toolbars.
chat-messageRole-based message container (user vs assistant) with distinct styling.
chat-message-areaMessage feed scroll layout with automatic message stacking.
message-scrollerSticky auto-scrolling viewport with pinned bottom stream anchor.
conversationComplete chat conversation feed with empty states and scroll buttons.
bubbleChat speech bubble with sentiment reactions and tail variants.
shimmerLoading placeholder shimmer text animation during LLM token streaming.
open-in-chatQuick-launch menu to transfer prompts directly to ChatGPT, Claude, or Perplexity.

3. Reasoning, Tools & Autonomous Agents

ComponentDescription
reasoningCollapsible chain-of-thought accordion with live thought duration timer.
chain-of-thoughtStep-by-step reasoning timeline with search queries and image badges.
toolAgent tool call inspection card with dynamic inputs, outputs, and execution states.
agentVisual agent profile card with instructions, registered tool manifests, and schema outputs.
planCollapsible execution plan breakdown with active progress shimmers.
taskMulti-step agent task list with file references and progress counters.
confirmationHuman-in-the-loop approval dialog for dangerous actions and migrations.
personaAssistant persona profile card with avatar, role, and capabilities.
queueBackground job queue and generation task progress monitor.

4. Code & Developer Workflows

ComponentDescription
code-blockSyntax-highlighted code block powered by Shiki with language picker and one-click copy.
sandboxInteractive code execution sandbox with console output capture.
terminalSimulated CLI terminal window with ANSI colored output and streaming cursor.
test-resultsTest suite breakdown card with pass/fail summary and execution duration.
stack-traceFormatted exception stack trace analyzer with clickable file frame traces.
commitGit commit change inspector with visual added/removed diff badges.
checkpointState snapshot version tracker with one-click restore functionality.
snippetCompact code snippet card with quick copy button.
schema-displayInteractive JSON Schema tree explorer and schema validator.
file-treeInteractive folder/file directory hierarchy browser.
environment-variablesKey-value environment variable manager with secret masking.

5. Context, Artifacts & Canvas

ComponentDescription
artifactSidecar panel for displaying generated documents, code artifacts, and previews.
contextLLM token usage counter, window limits, and prompt cache hit rate monitor.
model-selectorLLM model picker with provider logos (Anthropic, OpenAI, Google, Ollama) and keyboard shortcuts.
sourcesCollapsible RAG knowledge base sources panel with reference badges.
inline-citationPer-sentence citation footnote with hover card preview.
canvasInfinite 2D workspace surface for AI workflow visual graphs.
nodeWorkflow canvas node card with ports and actions.
edgeVisual edge connector connecting workflow nodes.
connectionNode connection routing line for AI graph workflows.
controlsZoom, pan, and fit-to-view navigation controls for canvas workflows.
panelFloating sidecar tool panel for canvas workspaces.
toolbarFloating contextual action toolbar for text selection and canvas items.

6. Media & Rich Previews

ComponentDescription
attachmentsMulti-file attachment carousel and thumbnail hover preview grid.
attachmentSingle file attachment badge with upload progress indicators.
audio-playerAI speech and voice synthesis audio playback controller powered by MediaChrome.
transcriptionSynchronized audio transcript viewer with timestamps and active segment highlighting.
jsx-previewSafe JSX string sandbox renderer for live AI UI generation.
web-previewResponsive viewport browser frame for previewing generated web pages.
imageAI generated image viewer with lightbox magnification and metadata.
package-infoPackage manifest and npm dependency analyzer.