cloud-monitoring-alert-configuration

Configure Cloud Monitoring alerting policies using Terraform and PromQL across Google Cloud resources.

23.7k|2.0k|Updated Mar 31, 2026
Ready to run
npx skills add https://github.com/google/skills --skill cloud-monitoring-alert-configuration-google
💬 Prompt for ChatGPT / Claude Chat▼
Please help me install this Agent Skill.
Skill: cloud-monitoring-alert-configuration
Source: https://github.com/google/skills/tree/main/skills/cloud/cloud-monitoring-alert-configuration
Command: npx skills add https://github.com/google/skills --skill cloud-monitoring-alert-configuration-google

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Manual alerting policy setup across Cloud Run, GKE, Compute Engine, and AI agents is error-prone, inconsistent, and time-consuming. This Skill automates the creation of production-grade alerting policies with statistical rigor, proper thresholds, and operational best practices baked in. ## Core Features & Use Cases - Golden Signals Coverage: Automatically generates latency, traffic, error, and saturation alerts for your workload. - PromQL & SQL Queries: Handles both Prometheus Query Language for metrics and SQL for trace-based reliability analysis. - Resource-Specific Routing: Detects your target (Cloud Run service, GKE cluster, AI agent, Compute Engine VM) and applies the correct alert templates. - Use Case: Your Cloud Run service is experiencing occasional latency spikes. Use this Skill to generate a Z-score-based latency alert, error rate SLO burn-rate policies, and model/tool call error tracking—all parameterized and ready to deploy. ## Quick Start Generate alerting policies for a Cloud Run service by specifying the service name and project ID.

Frequently Asked Questions about cloud-monitoring-alert-configuration

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I configure latency alerts for different traffic patterns?▼

This Skill analyzes your agent or workload traffic pattern (steady, seasonal, or bursty) and applies the appropriate algorithm: Long-Window Z-Score for stable traffic, Seasonal Decomposition for cyclical patterns, or Moving Averages for volatile workloads. Each uses different historical baselines to avoid false alerts.

What is the difference between Fast Burn and Slow Burn SLO alerts?▼

Fast Burn (1h and 5m windows, factor 14.4) detects urgent outages within minutes using a 5-minute buffer. Slow Burn (3d and 6h windows, factor 1.0) flags persistent degradation over hours or days without duration buffering because the long lookback prevents false positives.

Can I monitor model and tool call errors from my AI agent?▼

Yes. This Skill generates SQL-based alert policies that query OpenTelemetry trace spans in Observability Analytics, extracting error rates for individual models and tools. Requires telemetry enablement and a linked BigQuery dataset for the Trace scope.

Does this Skill support quality and safety alerts for agents?▼

Yes, through companion policies: Quality alerts monitor hallucination and response quality scores from Online Monitors; Safety alerts track Model Armor policy violations; Security alerts detect high IAM permission denied rates in audit logs. Each requires separate enablement and billing approval.

What happens if my agent is brand new with no historical metrics?▼

The Skill asks you to specify your expected traffic pattern (steady, seasonal, or bursty) and applies default Short-Window Z-Score (1-hour baseline) to enable rapid alert activation. Once 1 week of history accumulates, you can switch to more sophisticated algorithms.

Why do I need notification channels attached to alerting policies?▼

Alerting policies without notification channels still incur Cloud Monitoring charges but won't notify anyone during incidents. This Skill warns you when no channels are attached and asks whether you want to configure email, PagerDuty, Pub/Sub, or Slack notifications.