One analytics model for ingestion and reporting
Ingest game, website and server events through the collector, and read project-scoped reports through the authenticated API. Events and profiles stay in Tinybird and application settings in Supabase.
How to install ↗Keep write and read credentials separate
The Publishable key authorizes collection for a registered site. Authenticated report reads use the application’s access controls. MCP uses a private API key, and backends send server events with a private Secret key; these credentials are not interchangeable.
Reuse the event envelope
Page and track payloads include a message ID, event timestamp, context, scalar properties, and optional user ID. The collector resolves the project from the Publishable key and derives anonymous identity instead of trusting a browser-supplied visitor ID.
Query the same definitions
KPI, time-series, pages, sources, activity, profile and funnel endpoints are implemented over Tinybird. Date ranges and filters should match the dashboard. SQL investigation is a separate bounded read path.
Operate your own pipeline
The app repository contains the API, worker, queue infrastructure, and Tinybird project. There is not yet a general public API-key REST contract for every hosted report. Self-hosted operators must configure origins, credentials, and resource limits.
Frequently asked questions
Can I read analytics with the Publishable key?
No. The Publishable key is for ingestion only.
Where does analytics data live?
Tinybird stores analytics events and materialized profiles. Supabase holds application state.
Is there a stable public read API for every report?
The app API endpoints exist, but hosted external REST authentication and rate-limit documentation are still being finalized. MCP provides the documented project-key read path.
Start with a clear measurement model.
Install the SDK, then explore traffic and product activity with the same definitions.