Orchestrating AI Agents for Maximum Impact
_{✦ An interactive presentation on building with Claude ✦}

What Is This? • Quick Start • The Workflow • Controls • Tech Stack

An animated presentation showcasing context engineering — the art of orchestrating AI agents through structured planning, parallel execution, and persistent memory.

The core insight: Don't just prompt. Engineer the context.

Brainstorm → Research Swarm → Deep Dives → Decisions → Track → Ship

Based on: A real case study bootstrapping a complex voice assistant with 20+ parallel agents, 175K lines of research, and zero rework commits.

# Clone and install
git clone https://github.com/hyperb1iss/context-engineering-demo
cd context-engineering-demo
pnpm install

# Start development
pnpm dev

# Build for production
pnpm build

Open http://localhost:5678 and press Space to begin.

src/
├── components/       # ConversationPlayer, AgentSwarm, StatCounter...
├── slides/           # Individual slide components
├── data/             # Conversation mock data
├── hooks/            # useKeyboardShortcuts, useFullscreen
├── lib/              # animations.ts, theme.ts
└── styles/           # Tailwind + custom CSS

This presentation about context engineering was itself built using context engineering:

1. Mined conversation history from the Haven project using Claude Code
2. Synthesized 85KB case study from real agent sessions
3. Created detailed specs (BUILD.md, DESIGN.md, SLIDES.md)
4. Handed off to Claude Code with full context
5. Parallel agents built components while Sibyl tracked the work

It's context engineering all the way down.

Created by Stefanie Jane

_{✦ Built with context engineering, naturally ✦}