dt-obs-hosts

This skill equips an agent to query Dynatrace host and process infrastructure telemetry using DQL (Dynatrace Query Language). It solves the problem of turning natural-language infrastructure questions (CPU usage, memory utilization, disk space, top processes, host discovery) into correct read-only Grail/Smartscape queries, and provides curated query patterns, metric names, and alert thresholds so results feed cleanly into analytical workflows like anomaly detection, forecasting, and seasonality analysis.

It centers on host and process metrics (dt.host.cpu.*, dt.host.memory.*, dt.host.disk.*, dt.host.net.*, dt.process.* families), inventory attributes (OS type, cloud provider, technology stack, versions), cost dimensions, and data-quality checks. Core workflows include host discovery and classification, resource-utilization monitoring against documented thresholds (80% warning, 90% critical), process troubleshooting, technology/database discovery, Kubernetes version distribution and version-skew detection for upgrade planning, cost analysis by cost center, security-oriented port and process inventory, EOL-version compliance, and rightsizing. Detailed reference files cover host metrics, process monitoring, container monitoring, and inventory discovery, with explicit routing to sibling skills for Kubernetes workloads, AWS inventory, service metrics, and cross-source topology joins.

It targets SREs, platform and infrastructure engineers, and observability practitioners already using Dynatrace who want reliable DQL for infrastructure health, capacity planning, compliance, and cost work. All example queries are read-only telemetry reads with no mutation or destructive operations.