Read a URL in Novus Learn: turn any web page into a cited project

Open /read and paste a link. The best candidates are pages that are stable, substantive, and self-contained — a documentation page, a standards document, a long explainer, a research summary. Novus Learn will read what is there, so a page that is mostly navigation, mostly comments, or mostly a cookie banner produces a thin project, and it will be honest about that rather than padding it out.

Think of the URL the way you would think of a citation you are about to rely on. If you would not want to defend the page in a footnote, extracting study material from it will not improve it. The tool makes the source visible; it does not make a weak source stronger.

Once the page is read, the project opens with the claims Novus Learn found. Read this list before you read anything downstream, because everything else is built from it. You are checking two things: that the claims are actually claims the page makes, and that nothing important is missing because it lived in an image, a table, or a script-rendered region the reader could not see.

This is the step people skip, and it is the one that pays. Every artifact after this inherits whatever the extraction got right or wrong, so five seconds spent scanning the claim list is worth more than five minutes spent puzzling over a flashcard that seems slightly off.

From there the project behaves exactly like a Wikipedia project: a summary that reorganises without adding, structured study notes, flashcards, a quiz, and a concept map. Each carries the citation back to the passage it came from, so a card you cannot quite believe is one click from the sentence that produced it.

For a technical page, the concept map is usually the most revealing artifact — documentation tends to describe relationships one paragraph at a time, and seeing them all at once often exposes a dependency you had not noticed. For a report, the notes and quiz tend to be the useful pair.

Some sources are not a page, they are a site. API documentation, a multi-part guide, a handbook — reading these one URL at a time means you never see how the parts fit. The /site surface indexes a whole site into a single project so the concept map spans the material rather than a slice of it.

Use it deliberately rather than by default. A site index is a bigger, slower, broader thing than a page read, and for a source you only need one section of, /read is the better tool. The rule of thumb: if you would open five pages from the same domain in five tabs, index the site instead.

Projects are saved locally in versioned IndexedDB snapshots on your device, and they show up in My Library at /library. That means the project persists, but it also means it lives in this browser and nowhere else — clearing browser data clears it, and another device will not have it.

A web page is also less stable than a pinned Wikipedia revision. Sites get rewritten, reorganised, and taken down. If the page you are reading matters, note where it came from and when, and treat the project as a study aid rather than as an archive of the source.

The last habit is the same one that governs every surface in Novus Learn: check the parts you intend to lean on. Follow a claim back to its sentence on the original page, and satisfy yourself that the extraction preserved the meaning and not just the words. Extractive tools do not hallucinate, but a sentence pulled out of its qualifying context can still mislead.

Where the source is genuinely silent, the project will say so — and that silence is information. A page that cannot answer your question has told you to go find a better page, which is a more useful outcome than a confident paragraph that was never in any source at all.