1 · Discover

Sparke finds every same-origin link on the page, automatically.

When Sparke starts, it scans the current document for same-origin <a href> links – the pages you could navigate to next. There's no manifest and no annotation step: your ordinary HTML links are the route map.

What counts as a link

A link is eligible for preloading and interception when it points to a same-origin http/https page. Sparke deliberately skips:

  • links with a target other than _self
  • download links and rel="external"
  • hash-only links, and mailto: / tel: / javascript:
  • anything you opt out with data-ignore

Because discovery is just a DOM scan, links added later – by a fetched page or by your own JS – are picked up too.

Then: preload

Once links are discovered, Sparke queues them for preloading while the browser is idle. That's the next move.