Teardown is a Chrome extension that pulls any web page apart in one click and tells you what is actually wrong with it. It is built for the moment you are sat on a client site, a competitor's site or a staging build and you want the answer now, not after a crawler has finished its run.
Most heading tools print an H1 to H6 list and stop there. Teardown opens with the question that decides everything else: can this page be indexed at all? It reads the robots meta tag, canonical, hreflang and page language first, then works down through structure, metadata, images, content and design. It is live on the Chrome Web Store at version 1.7, and I designed, built and shipped all of it, extension, landing site, documentation and store listing.
picture srcset markup and video poster frames, plus alt text coverage across the lot.| Component | How it works |
|---|---|
| Heading outline | Headings are read from rendered text rather than raw markup, so pages built with character-split animation still come back as readable words instead of one long run-on string. |
| Keyword extraction | A TreeWalker walks every text node, links and bold text included, because that is exactly where commercial terms tend to sit. Tokenising uses Unicode property escapes with Intl.Segmenter for scripts that do not put spaces between words, and stop words are chosen from the page's own lang attribute. |
| Content scoping | The content root has to earn it. A lone article must hold at least a third of the page's text before it is trusted, which keeps card grids and carousel slides from being mistaken for the page body. |
| Scoring model | Four weighted categories, with a noindex page capped at 40 no matter how well it scores elsewhere. Nothing scores for merely existing, and nothing is scored off an unreliable measurement, so page weight is reported as a diagnostic instead. |
| Scan comparison | Outlines are compared with a longest common subsequence diff, so inserting one heading flags that one heading rather than everything below it. |
| Privacy model | No account and no backend. History lives in chrome.storage.local and goes when you uninstall. Content scripts inject only after you click Scan, so the extension cannot see a page until you ask it to. |