Observability Service Health

This connects Claude to Elastic's APM and Observability APIs so it can assess service health for you. It checks SLOs, firing alerts, ML anomalies, throughput, latency percentiles, error rates, and dependency health using ES|QL queries against APM indices. Honestly most useful when you're triaging production issues and want an AI to correlate multiple health signals instead of clicking through Kibana dashboards yourself. It pulls from the Kibana repo for APM-specific logic, which means it understands the domain model rather than just blindly querying metrics. From Elastic's official agent skills collection, so the integration patterns should be solid.