Skip to main content

​
Overview

Use task(name: str, method_name: Optional[str] = None) to mark a task inside a workflow. Tasks can wrap LLM calls, tool executions, or custom steps.

​
Function usage

from keywordsai_tracing.decorators import workflow, task

@workflow(name="math_workflow")
def math_workflow():
    @task(name="add")
    def add(a, b):
        return a + b

    @task(name="square")
    def square(x):
        return x * x

    return square(add(2, 3))

print(math_workflow())

​
Class usage

from keywordsai_tracing.decorators import workflow, task

@workflow(name="agent_run", method_name="run")
class Agent:
    @task(name="respond")
    def respond(self, prompt: str):
        return f"Echo: {prompt}"

    def run(self):
        return self.respond("Hello")

print(Agent().run())

​
Parameters

  • name: task display name
  • method_name (optional): required when decorating a class method to indicate the entrypoint

​
Tips

  • Keep tasks small and focused
  • Use task names to reflect intent, not implementation details