Мониторинг дрейфа метрик и KPI

★ 7.6 · vibe-coding

prd-v08-drift-baseline-compare is a Claude Code skill that implements a baseline → snapshot → compare → history monitoring pattern for any metric, KPI, or configuration that can change gradually over time. Where threshold-based alerting answers "is X over the line right now?", drift monitoring answers "is X different from last week's normal?" — catching slow regressions that fixed thresholds miss entirely. The skill captures a dated baseline, defines tiered warn/critical thresholds, sets a compare cadence (hourly to monthly depending on criticality), maintains append-only history logs at `status/drift-history/<id>.jsonl`, and generates compare scripts under `scripts/drift/`. Practical targets include AI search positions in ChatGPT and Perplexity, feature-flag rollout percentages, p95 latency per endpoint, per-user infrastructure cost, and per-channel CAC trends — anything where the direction of change matters more than the absolute value.