telemetry-queries

Query Prometheus for pod restarts and OOM events by collecting firing alerts and operational metrics from the Kubernetes namespace. This skill retrieves those signals to identify recent platform failures.

Analyzing Victoria Logs for platform errors involves searching structured logs and aggregating error volumes before retrieving detailed failure samples. This method prevents overwhelming the analysis context with raw logs during investigations.

Yes, you can monitor LLM cost metrics and NATS connection health with telemetry queries. The skill collects these health signals from the platform namespace to validate service health during observability investigations.

Identify failed knight tasks during a Night Watch sweep by querying observability systems for recent task failures and service health signals. This allows operators to find the highest-impact platform issues without parsing raw logs.

Yes, this approach works for collecting the last 24 hours of infrastructure failures. Use the skill to gather recent alerts, restarts, OOM events, and task failures, then summarize the most important issues found.