ai-agents-structured-response-each-task-execution-needs-to-return-a-machine-read-ab9eb68a

THEN Define a standard JSON response object with fields: 'status' (success/failed), 'duration', 'data_source', 'summary', and on failure 'error_category' (e.g., 'fetch_failed', 'db_missing', 'notify_failed'). Ensure all execution paths produce this structure. Log the response for observability.

# Pseudocode for structured result
import time, json

def execute_task():
    start = time.time()
    try:
        data = fetch_data()  # may raise
        result = {
            'status': 'success',
            'duration': time.time() - start,
            'data_source': 'api_x',
            'summary': 'Fetched 10 items'
        }
    except FetchError as e:
        result = {
            'status': 'failed',
            'duration': time.time() - start,
            'data_source': 'api_x',
            'summary': '',
            'error_category': 'fetch_failed',
            'error_detail': str(e)
        }
    # always output
    return result