Quickstart

Agent system for Oridecon — AI agents with tools, strategies, and execution.

Install

uv add oridecon-ai-agents

Minimal Agent

import asyncio

from oridecon import Application
from oridecon.ai.agents import AgentBase, AgentConfig, tool
from oridecon.ai.agents import AgentsModule
from oridecon.contracts.ai import AgentExecutorProtocol


@tool
async def get_weather(city: str) -> str:
    """Get the current weather for a city."""
    return f"Sunny, 22°C in {city}"


class WeatherAgent(AgentBase):
    name = "weather_agent"
    system_prompt = "You are a helpful weather assistant."

    @property
    def tools(self):
        return [get_weather]


async def main():
    config = AgentConfig(max_iterations=10, default_temperature=0.3)
    async with Application.boot(
        name="agent-demo",
        modules=[AgentsModule.configure(config)],
    ) as app:
        executor = await app.container.resolve(AgentExecutorProtocol)
        result = await executor.run(
            agent=WeatherAgent(),
            message="What is the weather in Tokyo?",
        )
        if result.is_ok():
            print(result.unwrap().message)


asyncio.run(main())

What Just Happened

@tool decorated a function — auto-generated its JSON schema for the LLM.
AgentBase subclass declared identity, persona, and tools.
AgentsModule.configure(config) wired AgentsProvider into the container.
AgentExecutorProtocol resolved the executor from DI.
executor.run() returned Result[AgentResponse, AgentError].

Next Steps

Guide — agents, strategies, execution in depth
How-Tos — common agent patterns
Configuration — all config keys