Privacy and usage analytics

What we measure

Basic measurements count visits, feature use, navigation categories and form outcomes. A consistent 10% pseudonymous browser sample adds sessions, active time, return visits, attribution categories, conversion funnels, browser performance, public model popularity and coarse click heatmaps.

We do not collect raw searches, typed form contents, webhook URLs, error messages or stacks, screen recordings, full referrer URLs, or private model identifiers. Public catalog identifiers are validated on the server before aggregation. Traffic and campaign labels use fixed categories; unknown labels become “other.”

Storage and retention

A random identifier in browser local storage expires after 90 days. A session identifier uses session storage and renews after 30 minutes of inactivity. An opt-out preference cookie tells the server not to count future page visits. It contains no identifier. These analytics identifiers are hashed with a server secret before storage. Detailed sessions are retained for 30 days; aggregate interaction counts for 90 days. Daily visitor hashes expire after two UTC days. Raw IP addresses are not stored in these analytics tables.

Your watchlist and theme are separate browser preferences. Alert subscriptions and source suggestions store the information you submit to provide those services. Infrastructure access logs are separate from product analytics.

Private messages

Contact messages and issue reports store the details you submit, including your reply address, in a private maintainer inbox. They are delivered to the maintainer’s private Discord channel and are never published or included in usage analytics. Human verification and submission limits help prevent spam.

Optional usage analytics

Usage analytics are enabled in this browser. Global Privacy Control and Do Not Track also disable interaction tracking.