example

Durable Agents Example

This example demonstrates the Convex Durable Agents component with a chat interface.

Features

• Thread-based conversations with AI
• Tool execution (weather lookup)
• Real-time streaming responses
• Status indicators and controls (stop, retry)
• Thread management (create, delete, list)

Setup

1. Install dependencies:

npm install

2. Set up your Convex project:

npx convex dev

3. Configure your AI model:

Edit convex/chat.ts and replace the mock model with your actual AI model:

// Example with OpenAI:
import { openai } from "@ai-sdk/openai";
const model = openai("gpt-4o");

// Example with Anthropic:
import { anthropic } from "@ai-sdk/anthropic";
const model = anthropic("claude-sonnet-4-20250514");

4. Set your API keys in the Convex dashboard or .env.local:

OPENAI_API_KEY=your-key-here
# or
ANTHROPIC_API_KEY=your-key-here

5. Run the development server:

npm run dev

Project Structure

example/
├── convex/
│   ├── chat.ts          # Agent definition and API
│   ├── tools/
│   │   └── weather.ts   # Tool implementations
│   ├── schema.ts        # App schema (empty - component manages tables)
│   └── convex.config.ts # Component registration
└── src/
    ├── App.tsx          # Chat UI
    └── main.tsx         # Entry point

Usage

1. Click "New Chat" to start a conversation
2. Ask about the weather: "What's the weather in San Francisco?"
3. The agent will use the weather tool to fetch data
4. Watch the streaming response and tool execution status
5. Use Stop/Retry buttons to control the conversation

Available Tools

• get_weather - Returns weather conditions and temperature for a city
• get_temperature - Async tool demonstrating delayed results

Notes

• The weather tool returns mock data for demonstration
• In production, replace with actual API calls
• The component handles all durability concerns automatically