Defining Agents

Decorator-Based Agents

from astromesh_adk import agent

@agent(name="assistant", model="openai/gpt-4o", description="General assistant")
async def assistant(ctx):
    """You are a helpful assistant."""
    return None  # Use default orchestration

The docstring becomes the system prompt. Return None to use the default orchestration pipeline, or use ctx.run_default() for pre/post processing.

Class-Based Agents

from astromesh_adk import Agent, RunContext

class MyAgent(Agent):
    name = "custom"
    model = "ollama/llama3"
    pattern = "plan_and_execute"

    def system_prompt_fn(self, ctx: RunContext) -> str:
        return f"You help {ctx.user_id}"

    async def on_before_run(self, ctx: RunContext):
        """Called before execution."""

    async def on_after_run(self, ctx: RunContext, result):
        """Called after execution."""

    async def on_tool_call(self, ctx: RunContext, tool_name: str, args: dict):
        """Called before each tool call."""

Configuration Options

Parameter	Type	Description
name	str	Agent identifier
model	str	Provider/model (e.g., "openai/gpt-4o")
tools	list	Tools available to the agent
pattern	str	Orchestration: react, plan_and_execute, parallel, pipeline, supervisor, swarm
memory	str | dict	Memory config (e.g., "sqlite")
guardrails	dict	Input/output guardrails
fallback_model	str	Fallback provider/model
routing	str	Routing strategy