sub-agents

Multi-Specialist Agent (Sub-Agents)

Demonstrates how to deploy multiple specialist agents from a single service using the app/agents/ directory. Each file in app/agents/ becomes an independent A2A endpoint alongside the main app/agent.ts coordinator.

This example has three agents:

• Coordinator (app/agent.ts) — routes users to the right specialist
• Math specialist (app/agents/math.ts) — performs arithmetic using the calculate tool
• Text specialist (app/agents/text.ts) — analyzes text using the word-count tool

Quick Start

bun install

# Create .env.local with your keys
echo "OPENAI_API_KEY=sk-..." > .env.local

bun run dev

Project Structure

app/
├── agent.ts            # Coordinator → /agent
├── agents/
│   ├── math.ts         # Math sub-agent → /math/agent
│   └── text.ts         # Text sub-agent → /text/agent
├── tools/
│   ├── calculate.ts    # Arithmetic (add/subtract/multiply/divide)
│   └── word-count.ts   # Text analysis (words/characters/sentences)
└── icon.png

API Endpoints

A single deployment exposes three independent A2A endpoints:

Endpoint	Description
/.well-known/agent-card.json	Coordinator discovery card
POST /agent	Coordinator JSON-RPC
/math/.well-known/agent-card.json	Math agent discovery card
POST /math/agent	Math agent JSON-RPC
/text/.well-known/agent-card.json	Text agent discovery card
POST /text/agent	Text agent JSON-RPC
POST /mcp	Shared MCP tool endpoint

Sub-Agent Pattern

Each file in app/agents/ follows the same shape as app/agent.ts. The filename determines the URL prefix:

// app/agents/math.ts  →  /math/agent
export const accepts: Accepts = { scheme: "exact", price: "$0.001" };
export default new ToolLoopAgent({ ... });

Environment Variables

Variable	Description
OPENAI_API_KEY	OpenAI API key

Payment

Each agent charges $0.001 per request via x402 on Base (mainnet in production, Base Sepolia in development).

Build & Deploy

bun run build
vercel