SignalFlow is the Python-like streaming analytics language behind every chart and detector in Splunk Observability Cloud.
Every SignalFlow program starts by selecting a metric stream with data(metric_name, filter=...), optionally narrows the stream by dimensions, custom properties, or tags using filter() combined with and, or, or not, and outputs the resulting stream with .publish(label=...). Without .publish(), a SignalFlow stream is computed internally (for intermediate variables) but never plotted on a chart or evaluated by a detector.
Even when you build a chart using the Splunk Infrastructure Monitoring Plot Builder UI, Splunk translates your clicks into SignalFlow under the hood—you can switch to the SignalFlow editor at any time to inspect or customize the exact code.
prod_filter = filter('env', 'prod') and not filter('host.name', 'test-*')
stream = data('cpu.utilization', filter=prod_filter)
stream.mean(by=['service.name']).publish(label='Avg Prod CPU')