malloy-analyze

Start by calling get_context to learn the available sources and fields, then profile columns with aggregate queries, form hypotheses, and investigate them with group_by and nested views. Validate each finding against a second metric before synthesizing results into a dashboard view.

Create a view annotated with # dashboard that nests KPI cards (# big_value), a time trend (# line_chart), and a category breakdown (# bar_chart). Reference existing measures from the base source rather than redefining them inline.

Malloy supports # big_value for KPIs, # line_chart for time trends, # bar_chart for category breakdowns, # dashboard for nested multi-visualization layouts, and # transpose for grouped summaries. Chart views render only the first aggregate, so use one aggregate per chart.

Notebooks inherit filters declared on the model source via #(filter) annotations; the publisher renders the widgets and injects where clauses server-side. The notebook-level ##(filters) syntax is unsupported legacy and should not be used.

Compile errors in .malloynb files are not surfaced by the linter; they only appear when a cell is executed. Run each malloy cell individually to catch syntax or reference errors during authoring.