An agent-powered creative platform built on Livepeer. Chat with AI to generate, edit, animate, and live-stream media — all on an infinite canvas.
Live: https://storyboard-rust.vercel.app
Type a prompt, and the agent orchestrates 40+ AI models to produce media on a visual canvas:
"Create a 6-scene story about a brave knight, animate the best scene,
and start a live stream with cyberpunk style"
The agent plans, selects models, runs inference in parallel, and places results as draggable cards connected by dependency arrows.
- Generate — images, videos, audio, 3D models from text prompts
- Edit — restyle, upscale, remove background, replace objects, LEGO/logo/isometric styles
- Animate — image → video via Seedance 2.0, Veo, LTX with automatic fallback chains
- Talking Video — generate speech + animate face (
/talk) - Live Stream — real-time LV2V with Scope pipeline + prompt traveling (
/stream) - Stories — 6-scene illustrated stories with style consistency (
/story) - Films — 4-shot mini-films with camera directions (
/film) - Projects — manage, replay, and organize creative projects (
/project) - Voice Clone — clone voice from audio sample for TTS
- Image Analysis — Gemini Vision extracts style/characters/mood (
/analyze) - Episode Groups — drag cards into episodes, context overrides
Type /help for all 25+ slash commands.
┌───────────────────────────────────────────────────────────┐
│ Your Application │
│ ┌─────────────────────────────────────────────────────┐ │
│ │ @livepeer/creative-kit │ │
│ │ ArtifactStore · ChatPanel · InfiniteBoard │ │
│ │ CommandRouter · CapabilityResolver · IntentClassifier│ │
│ └───────────────────────┬─────────────────────────────┘ │
│ │ │
│ ┌───────────────────────┴─────────────────────────────┐ │
│ │ @livepeer/agent │ │
│ │ AgentRunner · ToolRegistry · WorkingMemory │ │
│ │ GeminiProvider · ClaudeProvider · LivepeerProvider │ │
│ └───────────────────────┬─────────────────────────────┘ │
└──────────────────────────┼────────────────────────────────┘
│
┌──────────────────────────┴────────────────────────────────┐
│ Livepeer Infrastructure │
│ │
│ SDK Service (sdk.daydream.monster) │
│ /inference · /capabilities · /llm/chat │
│ /stream/start · /stream/publish · /stream/control │
│ │ │
│ BYOC Orchestrator (go-livepeer) │
│ 40 capabilities · fal.ai adapter · payment tickets │
│ │ │
│ Scope Orchestrators (LV2V) │
│ longlive pipeline · real-time video transform │
│ │ │
│ AI Model Providers │
│ fal.ai · Gemini · ByteDance · Tripo3D · xAI │
└────────────────────────────────────────────────────────────┘
Layer 0 — Agent SDK (@livepeer/agent)
Provider-agnostic agent runtime. AgentRunner manages the LLM ↔ Tool loop. Swap providers (Gemini, Claude, OpenAI, Livepeer) without changing app code. Working memory (800 token budget) + session memory (queryable logs).
Layer 1 — Creative Kit (@livepeer/creative-kit)
Reusable framework for creative AI apps. ArtifactStore (canvas state), ProjectPipeline (batch generation), CommandRouter (slash commands), CapabilityResolver (model selection + fallback chains), UI components (InfiniteBoard, ChatPanel, ArtifactCard).
Layer 2 — Applications
- Storyboard — Professional creative workspace (this app)
- Creative Lab — Educational app for kids 8-16 (
apps/creative-lab/)
| Provider | Route | API Key |
|---|---|---|
| Gemini (default) | /api/agent/gemini → Google API |
GEMINI_API_KEY |
| Claude | /api/agent/chat → Anthropic API |
ANTHROPIC_API_KEY |
| OpenAI | /api/agent/openai → OpenAI API |
OPENAI_API_KEY |
| Livepeer | /api/llm/chat → SDK → Gemini/Claude/OpenAI |
DAYDREAM_API_KEY only |
The Livepeer provider routes all LLM calls through Livepeer infrastructure — one API key for everything.
| Category | Models |
|---|---|
| Image | flux-dev, flux-schnell, recraft-v4, gemini-image, seedream-5-lite, nano-banana, flux-flex |
| Edit | kontext-edit, flux-fill |
| Video T2V | veo-t2v, ltx-t2v, pixverse-t2v |
| Video I2V | seedance-i2v, seedance-i2v-fast, veo-i2v, ltx-i2v, pixverse-i2v, kling-i2v |
| TTS | chatterbox-tts, gemini-tts, inworld-tts, grok-tts |
| 3D | tripo-t3d, tripo-i3d, tripo-mv3d |
| Other | bg-remove, topaz-upscale, talking-head, face-swap, lipsync, music, sfx |
All models have automatic fallback chains — if one rejects (content policy, size limit), the next sibling model tries automatically.
git clone https://github.com/livepeer/storyboard.git
cd storyboard
npm install
npm run dev # Storyboard on http://localhost:3000cd apps/creative-lab
npm run dev # Creative Lab on http://localhost:3001npm run dev # Development server
npm run build # Production build
npm test # Unit tests (481 tests)
npm run test:e2e # Playwright E2E testsClick the gear icon to configure:
- SDK Service URL —
https://sdk.daydream.monster(default) - API Key — Daydream API key for inference
storyboard/
├── packages/
│ ├── agent/ # @livepeer/agent — AgentRunner, providers, memory
│ ├── creative-kit/ # @livepeer/creative-kit — stores, routing, UI
│ ├── agent-pack-canvas/ # Canvas tools for agent
│ └── agent-pack-projects/ # Project tools for agent
├── app/ # Next.js App Router
│ ├── page.tsx # Main page + plugin registration
│ └── api/
│ ├── agent/ # LLM proxy routes (gemini, chat, openai)
│ ├── llm/chat/ # Livepeer unified LLM endpoint
│ ├── mcp/ # MCP proxy (discover, call, auth)
│ └── upload/ # GCS file upload
├── lib/
│ ├── agents/ # Agent plugins (gemini, claude, openai, livepeer)
│ ├── tools/ # 21 storyboard tools (create_media, scope_*, project_*)
│ ├── canvas/ # Canvas store + types (extends creative-kit Artifact)
│ ├── projects/ # Project store + /project commands
│ ├── story/ # /story generator + StoryCard
│ ├── film/ # /film generator + FilmCard
│ ├── stream-cmd/ # /stream planner + prompt traveling
│ ├── stream/ # LV2V session (scope params, graphs, frame extraction)
│ ├── layout/ # Layout engine (grid, narrative, episode, project grouping)
│ ├── skills/ # Slash command router + 23 skills
│ ├── sdk/ # SDK client (inference, capabilities)
│ └── mcp/ # MCP tool discovery + execution
├── components/
│ ├── canvas/ # InfiniteCanvas, Card, ContextMenu, CameraWidget
│ └── chat/ # ChatPanel, MessageBubble, StoryCard, FilmCard
├── apps/
│ └── creative-lab/ # Kids educational app (missions, gallery, safety)
├── skills/ # 23 skill files (.md) — agent system prompts
├── tests/
│ ├── unit/ # Vitest (481 tests)
│ └── e2e/ # Playwright
└── docs/
├── how-to-agent-sdk-arch.md # Architecture guide for building creative apps
└── superpowers/ # Design specs + implementation plans
| Document | What it covers |
|---|---|
docs/how-to-agent-sdk-arch.md |
How to build creative apps with Agent SDK + Creative Kit + Livepeer |
CLAUDE.md |
Full project instructions — infrastructure, capabilities, debugging |
docs/key-insights-scope.md |
LV2V debugging guide (Scope pipeline) |
docs/superpowers/specs/ |
Design specs for major features |
docs/superpowers/plans/ |
Implementation plans |
Copyright Livepeer Foundation.