Prometheus Metrics

When Prometheus is enabled, the Runtime API exports the following metrics:

Metric Type Description
runtime_uptime_seconds gauge Process uptime in seconds
runtime_project_uptime_seconds gauge Current project session uptime
runtime_project_loaded gauge 1 if a project is loaded, 0 otherwise
runtime_errors_total gauge Cumulative errors across all sessions
runtime_warnings_total gauge Cumulative warnings across all sessions
runtime_state gauge Numeric runtime state (0=Idle, 1=Loading, 2=Running, 3=Stopping)
runtime_compute_time_ms_short gauge Short-window average compute time
runtime_compute_time_ms_long gauge Long-window average compute time
runtime_congestions gauge Congestion count for current project
runtime_queue_flushes gauge Queue flush count for current project
runtime_api_calls_total counter Total API calls, labeled by endpoint
runtime_last_time_to_first_frame_seconds gauge Last measured time-to-first-frame