detectkit 0.52.0__tar.gz → 0.52.1__tar.gz

{detectkit-0.52.0/detectkit.egg-info → detectkit-0.52.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: detectkit
-Version: 0.52.0
+Version: 0.52.1
 Summary: Metric monitoring with automatic anomaly detection
 Author: detectkit team
 License: MIT

{detectkit-0.52.0 → detectkit-0.52.1}/detectkit/__init__.py

@@ -4,7 +4,7 @@ detectk - Anomaly Detection for Time-Series Metrics
 A Python library for data analysts and engineers to monitor metrics with automatic anomaly detection.
 """
 
-__version__ = "0.52.0"
+__version__ = "0.52.1"
 
 from detectkit.core.interval import Interval
 from detectkit.core.models import ColumnDefinition, TableModel

{detectkit-0.52.0 → detectkit-0.52.1}/detectkit/cli/assets/claude/rules/alerting.md

@@ -360,7 +360,7 @@ referenced by path). Key variables:
 | `{expected_range}` | one-sided-aware band (`>= 7.00`, `<= 1.10`, `[lo, hi]`, `N/A`) |
 | `{detector_name}`, `{detector_count}` | who fired (`"N detectors"` for multi) |
 | `{min_detectors}` / `{direction_policy}` / `{consecutive_required}` | the configured rule |
-| `{rule_display}` | full rule chip (legacy `min_detectors=… · direction=… · consecutive=…`, or also naming `anomaly_window`/`min_anomaly_share` when configured); recovery always renders the legacy form |
+| `{rule_display}` | full rule chip (legacy `min_detectors=… · direction=… · consecutive=…`, or also naming `anomaly_window`/`min_anomaly_share` when configured — on recovery too, so fire and recovery name one rule) |
 | `{window_points}` / `{window_matched}` | fraction-rule window size / matched count (empty unless configured / fired by it) |
 | `{direction}`, `{severity}` | observed values |
 | `{consecutive_count}` | **true** streak length (resolved at fire time, not capped at the rule) |

{detectkit-0.52.0 → detectkit-0.52.1}/detectkit/cli/assets/claude/rules/overview.md

@@ -68,6 +68,8 @@ separate from the `data_database` your queries read from.
 - `zscore` — mean/std; for clean, normally distributed data.
 - `iqr` — interquartile range; for skewed distributions / percentile metrics.
 - `manual_bounds` — fixed upper/lower thresholds (SLAs); no window, instant.
+- `autoreg` — prediction-based AR(p) on a trailing window; for fast-moving,
+  non-seasonal metrics and shape anomalies; stabilization on by default.
 
 `mad`, `zscore`, `iqr` share one windowed implementation, so they accept an
 identical parameter set (window, seasonality grouping, preprocessing, recency
@@ -78,7 +80,9 @@ weighting, detrending). See `detectors.md`.
 An **alert** is the primary entity; a detector anomaly is secondary evidence
 that a rule interprets. The rule is a per-point **quorum** (`min_detectors`
 detectors agreeing under a `direction` policy) that must hold for
-`consecutive_anomalies` grid-adjacent points. Notifications lead with the alert
+`consecutive_anomalies` grid-adjacent points — or, optionally, for a share of
+a trailing window (`anomaly_window` + `min_anomaly_share`, OR-ed with the
+consecutive rule). Notifications lead with the alert
 and the rule it fired on, with the anomaly shown as evidence. See
 `alerting.md`.
 

{detectkit-0.52.0 → detectkit-0.52.1/detectkit.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: detectkit
-Version: 0.52.0
+Version: 0.52.1
 Summary: Metric monitoring with automatic anomaly detection
 Author: detectkit team
 License: MIT