Generate an analytics dashboard with the data pipeline already wired

What you get in the generated ZIP

A typical analytics-dashboard generation includes:

app/blueprints/
├── metric_bp.py              # registered metric definitions
├── aggregate_bp.py           # materialised aggregate tables
├── dashboard_bp.py           # user-saved dashboards
├── widget_bp.py              # widget config (metric + chart type + filters)
├── filter_bp.py              # named filter sets per workspace
└── export_bp.py              # CSV / PDF export

app/models/
├── metric.py                 # name, sql_query, refresh_interval
├── aggregate_metric_daily.py # materialised daily rollup
├── aggregate_metric_hourly.py# materialised hourly rollup
└── dashboard.py              # user_id, workspace_id, layout JSON

app/services/
├── metric_refresh.py         # Celery Beat refreshes aggregates
└── query_executor.py         # workspace-scoped, parameterised, no SQL injection

What's already wired

• Materialised aggregates: every metric defines a refresh interval; the Celery Beat task wakes up and recomputes the aggregate table. Saves the production read replica from being hammered by dashboard queries.
• Workspace scoping: every query in query_executor.py injects the workspace_id filter. Cross-tenant data leakage is impossible by construction (no raw SQL outside the executor's allowlist).
• Time-series: hourly and daily rollup tables, with retention policies. Default 90 days hourly + 2 years daily; configurable per metric.
• Cohort analytics: users active in period N who were also active in period N-1 style queries are first-class citizens — there's a cohort_query() helper in query_executor.py.
• Frontend: Next.js + shadcn dashboard UI with Recharts. Drag-to-rearrange widget grid (using react-grid-layout). Filters apply across all widgets in a dashboard.
• Export: CSV from any widget, PDF of the entire dashboard via headless Chrome.
• Role-scoped data: viewers see only their workspace; admins see cross-workspace; superadmins see all (and every superadmin query is audit-logged).

What ships in docs/

• docs/decisions/ADR-0009-materialised-aggregates.md — chose materialised tables + scheduled refresh over real-time queries; rejected alternatives discussed
• docs/decisions/ADR-0013-time-series-retention.md — retention defaults and how to override
• docs/diagrams/data-pipeline.md — Mermaid diagram of source → aggregate → dashboard flow

Internal links

• PostHog integration for product analytics overlay
• for/vp-engineering covers reporting at organisational scale

CTA

Try it — free plan, no credit card. archiet.com.

Generate an analytics dashboard, look at the materialised-aggregate refresh and the workspace-scoped queries, decide if that's the data architecture you'd ship.