Profile your Amazon CloudWatch Logs estate with an AWS DevOps Agent skill in sandbox, no infrastructure to deploy

This article demonstrates how to build an AWS DevOps Agent skill that profiles your entire Amazon CloudWatch Logs estate in one view, showing log group classification, retention, usage patterns, subscription destinations, and estimated costs without requiring any standing infrastructure.

• Create a reusable skill bundling a Python analyzer script that runs in Sandbox during investigations
• Profile every log group across five dimensions: inventory, volume/cost, consumption, integrations, and optimization opportunities
• Identify optimization candidates like Standard-class groups with no metric filters or subscription filters suitable for Infrequent Access
• Detect passive log groups with no configured consumers and map subscription filter destinations
• Run analysis on-demand or schedule with EventBridge Scheduler and Lambda for regular cadence reviews
• Batch GetMetricData calls and run concurrent describe operations to minimize API round trips and wall-clock time

The skill treats your log estate as one system, correlating CloudWatch signals to present actionable options rather than mandates, enabling informed decisions about log group optimization and cost management at scale.