← ClaudeAtlas

technical-seolisted

What a page must declare to a crawler and why — metadata, canonical, structured data, crawlability. Use when building any page that will be public.
Deadshot-77/davinci · ★ 0 · AI & Automation · score 73
Install: claude install-skill Deadshot-77/davinci
# Technical SEO This covers what the markup declares. It does not cover keyword strategy, competitive positioning or content planning — those need search-volume data and business context you do not have, and producing them from nothing yields confident output with nothing behind it. **If a brief asks for keyword research, say plainly that it needs data outside the project rather than inventing it.** What you can do is make the page legible to a machine and honest about what it is. That part is verifiable, which is why it belongs here. ## Run the check on built output ``` node <plugin>/scripts/seo.mjs . ``` It reads `out/`, `dist/` or `build/` — not your source. A framework's metadata export, a layout's title, a component's alt attribute: none of them can be verified by reading the file they are written in, because what ships is the render. **If there is no build, the check does not run**, and it says so. That is not a pass, and reporting it as one is the same error as reporting a screenshot you never took. Errors are things a page cannot do without. Notes are judgement calls with context attached. Read both; fix the errors; decide about the notes. ## What every page owes **A `lang` on `<html>`.** Without it a screen reader guesses and mispronounces. **A `<title>` that is this page's, not the layout's.** The commonest defect the check finds is several routes sharing one title, because a layout set it and no route overrode it. The title is the name of the page in a tab