Skip to main content

@workflow Decorator

The @workflow decorator defines multi-step execution flows. Workflows orchestrate decisions, tools, and other workflows.

Basic Usage​

from waxell_sdk import agent, workflow

@agent(name="data-processor")
class DataProcessor:

    @workflow
    def process_data(self, ctx):
        # Step 1: Validate input
        validated = ctx.call(self.validate)

        # Step 2: Transform data
        transformed = ctx.call(self.transform, data=validated)

        # Step 3: Store results
        return ctx.call(self.store, data=transformed)

Execution Context​

The ctx parameter provides access to the execution context:

@workflow
def my_workflow(self, ctx):
    # Access input data
    input_data = ctx.input

    # Call other methods
    result = ctx.call(self.other_method)

    # Access configuration
    config = ctx.config

    return result

Durability​

Workflows support checkpoint and resume for long-running operations:

@workflow(durable=True)
def long_running_workflow(self, ctx):
    # Each ctx.call is a checkpoint
    step1 = ctx.call(self.expensive_operation)

    # If interrupted here, resumes from last checkpoint
    step2 = ctx.call(self.another_operation)

    return step2

Next Steps​