Your inputs stay
on your device.

Yarn quantities, measurements, generated models and copied shopping lists are processed in your browser. We do not upload those inputs or files, and we do not require an account.

What we count

On the two tool pages we send a small first-party event: which tool was opened, the UTC date, a broad referral category (such as search, a social site, direct or another website), and whether a shopping list was copied or an STL was generated and its browser download started. We cannot confirm that a file was saved or printed. A random ID lasts only for that page load and prevents a retry being counted twice. It is not a persistent visitor ID. We keep no names, IP addresses, raw referring URLs, filenames or form values in our metrics database.

Counts are not unique people. Bot filtering is imperfect, direct visits are not automatically organic growth, and missing referrers do not prove how someone found us.

Optional feedback

You can send one of three fixed choices: helpful, confusing, or something looks wrong. We record only the choice with the same anonymous page-load event information. There is no free-text box. A success message means the server accepted your feedback; a failure message means it was not confirmed.

Independent measurement

We also use Cloudflare Web Analytics to compare page-load totals independently. Cloudflare hosts this site and receives network information needed to deliver requests. See Cloudflare’s privacy policy for its processing. The analytics script is not loaded on this privacy page, the home page, the stats page, previews or visits marked for testing.

Controls

Do Not Track and Global Privacy Control signals disable both analytics and feedback sending in our client script. Adding ?qa=1 to a tool URL also disables them for testing. No cookies or persistent browser storage are used by our own code.

Public stats contain only daily grouped counts, never random navigation IDs. The JSON endpoint shows the last 28 UTC days. Anonymous event rows are retained for analysis; there is no automatic deletion schedule in this first version. Free-tier limits may make metrics or feedback temporarily unavailable; the calculators still work.

Who made this?

Diogo Neves, with substantial help from OpenAI Codex. These are small experiments, not certified engineering or yarn-suitability advice. The individual tools explain their limits and sources.