Downloads§

Every figure on this page is a sum of daily download counts from npm's public download statistics, the same counts npm shows on a package's own page. History is collected one day at a time rather than as a single all-time total, because npm answers a request spanning more than eighteen months by quietly returning only the most recent eighteen; long histories are fetched in bounded ranges instead, and every answer is checked against the package and the days it was asked for before anything is kept.

The daily records are committed to the repository, one file per package, and the site is built from those files without asking npm at build time or while being read. A refresh appends the days npm has counted since the last one, and re-reads the most recent 7 days because npm's latest counts can still settle; older days are never requested again. So a total is reproducible from the records at the commit a page was built from, and anyone can audit both the data and the collector.

A package's total is the sum of its stored days, and the ecosystem total is the sum of every published package's total; neither is adjusted for packages that depend on one another. Charts fold the daily records into days, weeks or months depending on how long a history is. The buckets are for reading; the stored records stay daily. Each package page links here with that package in focus, and the landing page carries the ecosystem total.

Thanks to the TanStack team for publishing their research into npm download statistics and laying much of the groundwork for this approach. Their write-up helped us avoid several npm API pitfalls, the silent eighteen-month truncation above all, and we are similarly keeping our methodology and data auditable. This is an acknowledgement of work they shared openly, not an association with or endorsement of HyperFrontend.