TheLinksMaster - SEO & AI SEO Toolkit by 24Bit System

TheLinksMaster - SEO & AI SEO Toolkit opens as a Firefox sidebar — click its toolbar icon, or press Ctrl+Shift+U, on the page you are already looking at, and it audits that page: 97 checks across indexing, on-page tags, content, images, links, structured data, performance, response headers, site files, rendering and AI visibility, scored into the same five weighted categories a professional audit uses (Technical, Performance, Content, Crawlability, AI visibility).
try now https://www.thelinksmaster.com/

It is local by design. There is no account, no API key, no telemetry and no analytics. The analysis happens in the tab: the extension transmits no page content and holds no host permissions, so it can only reach the site you are standing on — the same origin the browser already loaded. The one outbound thing it does is a navigation you start: a link you click to our website, which carries the page address in ordinary link text and nothing else.

What it looks at:

• Indexing — status, indexability, noindex in meta robots, googlebot and X-Robots-Tag, robots.txt rules that cover this path, canonical presence and agreement, hreflang syntax and self-reference.
• On-page — title and description by both characters and rendered pixel width, so truncation is measured the way Google truncates instead of the way a character counter guesses. Heading outline, level jumps, multiple H1s, doctype, charset, lang, viewport.
• Content — word count, text-to-HTML ratio, Flesch readability, the terms the page actually repeats, and whether that primary term reaches the title, H1 and description.
• Images — alt present versus alt empty (different bugs), width/height pairs and the layout shift they prevent, lazy-loading declarations, format and bytes.
• Links — internal versus external balance, nofollow coverage, generic and empty anchors, target=_blank without rel="noopener", insecure hrefs.
• Structured data — JSON-LD parsed for real, so a broken block is reported as a broken block rather than counted as markup present, and a block with no schema.org @context is reported as the missing line it is.
• Performance — Navigation Timing measured on your machine, resource and byte counts, DOM size and depth, render-blocking scripts, compression and caching headers.
• Site files — /robots.txt, the sitemaps it declares plus the common paths, <lastmod> freshness, /llms.txt, /ads.txt.
• Raw HTML versus rendered DOM — the strongest check here. A headless crawl is needed to learn this server-side, but your browser has already done the rendering: the panel compares the bytes the server sent with the document you are looking at and scores the signals that only exist after JavaScript runs. Google renders, most AI crawlers and link-preview bots do not.
• AI visibility — whether GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot, Bytespider, Applebot-Extended and the rest are blocked, plus semantic landmarks, answerability and structured data.

The panel is deliberately short. It is drawn as an instrument readout: the score in tabular mono on a 0–100 ruler with the rating thresholds marked on the scale, the five weighted categories as the same ruler with a notch at the 75 pass line, the critical/warning/passed counts, a confidence line, the three most urgent failing checks as a short "top issues" list, and a collapsed list of the checks a browser could not measure, each with its reason. Anything deeper belongs on the site, so the card carries one primary action: Get the full report opens the free SEO checker with this page's address and asks the site to start the audit on arrival, so the tab lands on the scored report; the site runs the server-side, crawl and site-wide checks a browser cannot see and stores the report there. More free tools opens the directory. Nothing opens unless you click, and no page content is transmitted — the address travels only as ordinary text in the link you chose to open.

Structured data that qualifies, not just parses. Anyone can add an Article block and get a green tick from a validator that only reads the JSON. The checks copy the properties out of each typed node and compare them with what Google documents per type: required for the few types that have them (Product, VideoObject, BreadcrumbList), recommended for the rest, plus the shape of each value — a date has to be ISO 8601, an image has to be an absolute URL, an offer has to carry a price. The comparison also notices types Google stopped turning into rich results, such as FAQPage (May 2026) and HowTo (2023), because the markup stays valid schema.org while the rich result does not. Nothing is graded until the node declares a schema.org @context, since that is what makes a block machine-readable at all: a typed block without one is told to you as a page that paid for markup search engines ignore, not as a page with no structured data. An @type Google has no property rules for is left unmeasured rather than passed, and the tables are maintained by hand from Google's own documentation, so they will drift as Google changes its features.

Deliberately not measured. Broken links and site-wide coverage need requests to other origins, so the panel marks them "not measured" instead of guessing, and points at the web tool. The same applies inside its own origin: if the panel cannot re-read the page's HTTP response, the status, security-header and compression checks report "not measured" rather than inventing a fault — or a pass. The browser's resource list is a third case: when a navigation exposes no timing entries, the subresource count is unknown, not zero. Core Web Vitals field data comes from CrUX, which is server-side. Rather than bury that, the summary names how many of the checks could not be measured on this page and why, each one in its own group.

Search-results reading (competitor titles, snippets and ranking domains from a results page you already have open) is off by default and switches on only in Settings. It parses the markup already on your screen and makes no request of its own.