Observability

Logging - JSON logs for lifecycle and events; avoid secrets

Metrics - Node: messages_total, errors_total, tick_duration - Edge: enqueued_total, dequeued_total, dropped_total, queue_depth, blocked_time - Scheduler: runnable_nodes, loop_latency, priority_applied

Tracing (optional) - Correlate via trace_id; spans around node and edge operations when enabled

Enable - Defaults are no-op; enable adapters via configuration flags or optional deps

Example: enable metrics (if a Prometheus adapter is available)

from meridian.observability.metrics import configure_metrics, PrometheusMetrics

# Create and register a metrics backend (noop by default if not configured)
configure_metrics(PrometheusMetrics(namespace="meridian"))