What problem does it solve? Answering questions like "how many Kiro credits were used yesterday in China time" requires querying the llm_gateway_usage_events LanceDB table, but LanceDB's query layer cannot reliably run aggregate expressions like sum(credit_usage). This Skill fetches the filtered rows and aggregates them locally in Python, producing accurate daily totals and per-account breakdowns. ## Core Features & Use Cases - Daily Credit Totals: Computes the global Kiro credit total for one Asia/Shanghai calendar day using a half-open time window. - Account Subset Breakdowns: Filters accounts by case-insensitive name substrings plus user-specified exact account names, with per-account totals and explicit reporting of exact accounts with zero usage. - Missing-Metering Awareness: Detects rows with missing credit_usage and reports totals as lower bounds instead of silently treating them as zero. - Use Case: Ask "昨天中国时间 Kiro 用了多少积分？账号名包含 laohan 的号用了多少？再把 KitWilliam 也算进去" and get the global total, subset total, and per-account breakdown in one report. ## Quick Start Use the kiro-usage-day-report skill to total yesterday's Kiro credits in China time and break down accounts containing laohan plus the exact account KitWilliam.
