Three new River-based plugins for InfluxDB 3 bring online machine learning directly into the write path: an Anomaly Detector, an Auto-Profiler, and a Forecaster. The Anomaly Detector supports rolling Z-score, seasonal, and ADWIN drift detection, maintaining per-series model state updated incrementally on every write. The Auto-Profiler acts as a control plane, profiling each numeric series and writing recommendations (pattern labels, detector modes, thresholds) that the Anomaly Detector consumes for adaptive tuning. The Forecaster uses River's SNARIMAX model to produce short-horizon, multi-step forecasts per series, self-throttling to evaluate full horizons before generating new ones. All three plugins run inside InfluxDB 3's embedded Python Processing Engine, write results back as queryable SQL tables, and require no external ML infrastructure or data pipelines.
