Reflect

Returns the catalog of Reflect cubes with their measures, dimensions, and joins, restricted to the members the current workspace and caller may actually query. Always call this first when writing a new analytics query so you use real measure/dimension titles instead of guessing. Filter by cube name to keep output small.

list_schema

Run a structured analytics query against the user’s Reflect workspace. Args: - query.measures: required array of fully-qualified measure titles, e.g. ["EMPLOYEES.COUNT"]. - query.dimensions: optional dimension titles, e.g. ["EMPLOYEES.DEPARTMENT"]. - query.dateRange / compareDateRange: { startDate, endDate } ISO dates. - query.filters: [{ member, operator, values }]. Cube operators: equals, notEquals, contains, gt, lt, gte, lte, set, notSet, inDateRange. - query.limit / offset / total. - query.order: [{ member, direction: "asc"|"desc" }, ...]. - scopeType: "COMPANY" (company-wide) or "USER_SCOPE" (filtered by caller’s permissions). Discover valid titles via list_schema first. Returns { data: row[], total? }.

get_query

Run a read-only SQL query against the user’s Reflect workspace via the Cube SQL API. Use this when the user asks for a complex query (custom aggregations, expressions, or joinsrankings, filtering on measures, etc.). Cube SQL rules: cubes are tables in UPPER_SNAKE_CASE, measures must be wrapped in MEASURE(), dimensions are selected directly, and every selected dimension must appear in GROUP BY. Discover valid cube and column names via list_schema first, and add a LIMIT to keep output small. scopeType: "COMPANY" (company-wide) or "USER_SCOPE" (filtered by caller’s permissions). Returns { rows: row[], rowCount }.

sql_query