What problem does it solve? Headquarters staff managing multiple restaurant locations need to know which stores to visit today, but querying deep insights for every store wastes API quota and produces noisy, misleading comparisons. This Skill triages all stores with a lightweight first pass, then deep-dives only the true outliers. ## Core Features & Use Cases - Two-pass triage: Runs a light business-summary sweep across all stores, then calls heavy store-insights only for the top N underperformers to protect rate limits. - Range-based outlier detection: Judges performance against each store's similar-business-day range (mean ± std) instead of naive week-over-week percentages, avoiding false alarms from event-driven spikes. - Deterministic scoring and handoff: Ranks stores with a fixed severity/direction formula, separates "no signal" from "insufficient data" pools, and hands off top stores to sales-review or sudden-drop-investigation skills. - Use Case: An HQ manager asks "which stores should I check this week?" and receives a prioritized table with each store's revenue deviation, a one-line hypothesis (traffic vs. ticket vs. menu mix), and one concrete action. ## Quick Start Ask the assistant to run a health check across all stores for this week and list which locations need attention first.
