fastadk 0.1.0

A developer-friendly framework for building AI agents with Google ADK

FastADK

FastADK is an open‑source Python framework that layers ergonomic abstractions over Google ADK and Vertex AI, enabling developers to design, test, and deploy tool‑using LLM agents with minimal boilerplate. Declarative decorators, auto‑generated FastAPI routes, pluggable memory back‑ends, and built‑in observability let teams prototype locally and scale the same code seamlessly to containerised, serverless environments.

Quality Assurance

Before submitting any code changes, please run our quality checks locally. See the Developer Guidelines for Quality Assurance for detailed instructions.

Features

• Declarative Agent Development: Create agents with simple @Agent and @tool decorators
• Configuration System: YAML/TOML-based configuration with environment variables support
• CLI Interface: Interactive command-line interface for agent development and testing
• Testing Utilities: Built-in testing framework for agent validation and simulation
• Comprehensive Observability: Logging, metrics, and tracing for monitoring agent behavior

Quick Start

from fastadk import Agent, BaseAgent, tool

@Agent(model="gemini-1.5-pro", description="Weather assistant")
class WeatherAgent(BaseAgent):
    @tool
    def get_weather(self, city: str) -> dict:
        """Fetch current weather for a city."""
        return {"city": city, "temp": "22°C", "condition": "sunny"}

# Run the agent
if __name__ == "__main__":
    import asyncio
    
    async def main():
        agent = WeatherAgent()
        response = await agent.run("What's the weather in London?")
        print(response)
    
    asyncio.run(main())

Configuration

FastADK supports configuration through YAML or TOML files and environment variables:

# fastadk.yaml
environment: dev

model:
  provider: gemini
  model_name: gemini-1.5-pro
  api_key_env_var: GEMINI_API_KEY

memory:
  backend_type: inmemory
  ttl_seconds: 3600

telemetry:
  log_level: debug

Testing

FastADK provides a testing framework for agent validation:

from fastadk.testing import AgentTest, test_scenario

class TestWeatherAgent(AgentTest):
    agent = WeatherAgent()
    
    @test_scenario("sunny_weather")
    async def test_sunny_weather(self):
        response = await self.agent.run("What's the weather in London?")
        assert "sunny" in response.lower()
        assert self.agent.tools_used == ["get_weather"]

Installation

pip install fastadk

License

FastADK is released under the MIT License.