Naklov SEO Audit — On-Page, Technical & AI Crawler Check

Naklov SEO Audit reads the page you are on and tells you what is wrong with it — in order of how much it matters. Most on-page tools hand you forty data points and leave you to work out which ones need attention. This one starts from the other end. Every check returns a severity and a plain sentence explaining the consequence. Not "noindex present", but "search engines will drop this page from their index entirely, so nothing else on this list matters until it is removed". WHAT IT CHECKS Indexing and technical · noindex in the meta tag and in the X-Robots-Tag response header · whether robots.txt blocks this path, HTTP status, redirect hops · HTTPS, canonical tags — missing, duplicated, or pointing elsewhere · hreflang sets, including the self-reference that is so often left out · sitemap discovery through robots.txt and the conventional location On-page · title and meta description, present and within displayable length · H1 count, heading outline, and levels that skip a step · word count, declared page language, viewport tag · images missing alt text, with decorative images excluded · the terms the page repeats, and whether the title covers them Structured data · JSON-LD with @graph flattening, and malformed blocks flagged · required properties per schema type — the silent reason rich results never appear · Microdata, RDFa, Open Graph and Twitter Card completeness Links · internal and external, nofollow, sponsored, ugc · broken internal links, found by actually requesting them · redirect chains, empty anchor text, target=_blank without rel=noopener WHO CAN READ YOUR PAGE — INCLUDING AI robots.txt now decides two separate things, and most sites treat them as one. Training crawlers such as GPTBot, ClaudeBot and Google-Extended feed model training. Answer agents such as OAI-SearchBot, PerplexityBot and Claude-SearchBot fetch pages to cite them in generated answers. Blocking the first group is a legitimate editorial choice. Blocking the second removes your page from AI search results — and it is usually accidental, inherited from a rule written years ago to opt out of training. The audit reads your robots.txt, evaluates ten named crawlers against this exact URL, and shows the two groups apart. Blocking a training crawler is never counted as a fault. It also checks for llms.txt. SPEED AND PREVIEW Core Web Vitals are read from this actual page load, not estimated: LCP, CLS and TTFB, alongside total page weight and the request breakdown. Images are measured too — file size, how much larger they are than the box they are drawn into, and missing width and height, which is the usual cause of layout shift. The preview tab shows the Google result as it will appear on desktop and mobile, truncated by pixel width the way Google does it rather than by character count, plus the social card for when the page is shared. IT REMEMBERS WHAT YOU FIXED Every audit is saved on your own device. Come back tomorrow and the popup shows what moved, what you fixed and what appeared since last time. A one-shot analyser tells you a page has four problems; a tool that remembers tells you two of them are new since Tuesday, which is the part you can act on. IT ASKS FOR NO SITE ACCESS This is the part worth reading twice. The extension requests activeTab, scripting and storage — and no host permissions at all. It cannot read any page you have not explicitly asked it to audit. That is possible because of how the checks are built. Response headers, robots.txt, llms.txt, the sitemap and internal link status codes are all fetched from the page's own origin, which needs no extra permission. External links are listed but never requested: checking them would mean asking for access to every website you visit, and that is a permission worth refusing. The trade-off is stated inside the extension, not hidden. NOTHING LEAVES YOUR COMPUTER The reading happens inside the page in your own browser. The verdict is computed locally. The history lives in chrome.storage on your device. There is no account, no server and no analytics. Export everything as JSON or delete it in one click. 26 LANGUAGES English, Turkish, German, French, Spanish, Italian, Portuguese (Brazil), Dutch, Polish, Russian, Ukrainian, Czech, Romanian, Greek, Swedish, Arabic, Hebrew, Persian, Hindi, Indonesian, Vietnamese, Thai, Japanese, Korean, Simplified and Traditional Chinese. Right-to-left layouts are properly mirrored, not just translated. The interface language can be set independently of your browser. Free. No account. No upsell.