alert inspector

The Alert Inspector analyzes how alerts are defined and triggered across a monitoring stack. It normalizes rule formats, evaluates thresholds and durations, and checks routing mappings to ensure alerts reach the right responders without duplication. It produces a structured report that highlights high-risk rules, potential false positives, and opportunities to reduce noise. Conceptually, it treats each alert as a rule with inputs (metric, threshold, duration, labels) and outputs (risk_score, recommended_changes, and a changelog). The tool is especially valuable after creating new alerts, consolidating alert groups, or during post-incident reviews to verify alignment with SRE practices. By modeling impact and collision potential, it helps teams design cleaner alert schemas and more precise on-call routing. Unique value arises from automated dedup detection, cross-rule dependency checks, and exportable change artifacts suitable for CI/governance. Use cases include evaluating Prometheus or CloudWatch rules, suggesting offsetting thresholds, and proposing runbook links for rapid remediation. Core features include rule analysis, threshold validation, duration consistency checks, dedup and collision detection, routing verification, and report export. The tool supports collaboration by delivering actionable recommendations without modifying live configurations directly.