Skip to main content

tinkr anomaly

Detect anomalies across all tracked metrics for a service. Compares current values against baselines and thresholds to surface statistically significant deviations.

tinkr anomaly <service> [options]

Arguments​

ArgumentDescription
serviceService name as configured in the active profile

Options​

FlagDefaultDescription
--since TEXT1hLook-back window — e.g. 30m, 2h, 6h
--severity TEXT—Filter results: low, medium, high, critical
--jsonoffEmit raw JSON

Examples​

# Check for anomalies in the last hour
tinkr anomaly payments-api

# Last 30 minutes, high and above only
tinkr anomaly payments-api --since 30m --severity high

# All severity levels
tinkr anomaly payments-api --since 2h

# JSON for alerting pipelines
tinkr anomaly payments-api --since 1h --json

Output​

Anomalies detected for payments-api (since 1h)

  HIGH     error_count       847 errors in 10m — threshold 10
  MEDIUM   latency_p99       3.2s — threshold 1s
  LOW      db_conn_pool      82% utilization (baseline: 45%)

Severity levels​

LevelMeaning
criticalService is effectively down; immediate action required
highStrong signal of an active incident
mediumDegradation detected; may escalate
lowMinor deviation; worth monitoring

Detection method​

The backend computes anomalies by comparing the current window against a historical baseline (typically 1–7 days of the same time-of-day). Thresholds can be tuned via alert rules — see tinkr alert.

Integration with watches​

tinkr watch runs anomaly detection on a schedule and fires notifications when the anomaly set changes. See tinkr watch.

See also​