Skip to content

Examples

Complete, runnable examples demonstrating AIDK features.

Running Examples

All examples are in the example/ directory.

bash
# Clone the repo
git clone https://github.com/rlindgren/aidk.git
cd aidk

# Install dependencies
pnpm install

# Run the backend
cd example
pnpm dev:backend

# In another terminal, run the frontend
cd example
pnpm dev:frontend

Example Structure

example/
├── backend/         # Express + AIDK server
│   ├── agents/      # Agent definitions
│   ├── tools/       # Tool implementations
│   ├── channels/    # Channel definitions
│   └── server.ts    # Express setup

├── frontend-react/  # React client
│   └── src/
│       ├── App.tsx
│       └── components/

└── frontend-angular/ # Angular client
    └── src/app/

By Feature

State Management

See how signals work:

  • Files: backend/agents/task-assistant.tsx
  • Features: comState, signal, computed

Tools

Custom tool definitions:

  • Files: backend/tools/calculator-tool.ts, backend/tools/scratchpad-tool.tsx
  • Features: Tool creation, execution, rendering

Channels

Real-time updates:

  • Files: backend/channels/todo-list.channel.ts, backend/routes/channels.ts
  • Features: Channel definition, publishing, subscribing

Client Integration

React

  • Files: frontend-react/src/App.tsx, frontend-react/src/hooks/
  • Hooks: useEngineClient, useExecution, useChannel

Angular

  • Files: frontend-angular/src/app/
  • Services: EngineService, ExecutionService, ChannelService

Persistence

Database integration:

  • Files: backend/persistence/
  • Features: Execution tracking, message logging, metrics

Next Steps

Explore a specific example:

Simple Chat

Basic chat agent with streaming responses.

Task Assistant

Full-stack app with tools, channels, and React.

Multi-Agent

Model-driven orchestration with Component Tools and Fork/Spawn.

Dynamic Router

Smart model selection based on task and cost.

User Memory

Tools that remember user context and personalize responses.

Voting Consensus

Multi-agent voting for near-zero error rates. Inspired by MAKER.

Progressive Adoption

Start simple, add features incrementally as needs grow.

Released under the MIT License.