SEO Audit Tool
100% private. 100% free. Your URL is fetched through our secure proxy and analyzed entirely in your browser. Nothing is stored, logged, or tracked.
SEO Checklist & AI Prompt
Use the prompt below with any AI coding agent to automatically implement every SEO, accessibility, and AI visibility best practice on your website.
What the audit actually measures
Paste an address and the tool pulls the page itself, its robots.txt, its sitemap.xml, its llms.txt and its stylesheets. Against those it evaluates 172 checks organised into seven categories: 27 for meta information, 17 for content structure, 42 technical, 35 for social and rich results, 13 for links, 21 for accessibility, and 17 for performance, three of which are Core Web Vitals.
How the score is built
Checks are not all equal: each carries a weight between 1 and 3. Passing earns the whole weight, a warning earns half, a failure earns nothing, and the ratio becomes your 0–100 score. The verdict bands are deliberately blunt: Excellent from 85 up, Fair from 65, Needs Work below that.
Where your URL goes: and doesn't
A small stateless proxy performs the fetch, because a browser cannot request arbitrary sites directly. It announces itself as SourceStrongAI-SEO-Audit/1.0, refuses localhost and private addresses, stops at 5 MB or 15 seconds, and stores nothing. Everything after that (parsing the DOM, evaluating all seven categories, computing the score) happens in your own browser. The one exception is measured openly: LCP, CLS, and TBT come from Google's public PageSpeed Insights API, requested straight from your browser, as lab numbers on the mobile profile rather than field data. No accounts, no saved reports.
Is GPTBot, ClaudeBot or PerplexityBot blocked by your robots.txt?
The audit reads your robots.txt the way a crawler does (a bot obeys the one group its own token matches, and ignores the rest) then reports whether each of 15 named AI agents can actually reach the site. GPTBot, OAI-SearchBot and ChatGPT-User for OpenAI; ClaudeBot, Claude-User and Claude-SearchBot for Anthropic; PerplexityBot and Perplexity-User; Googlebot, Google-Extended, Applebot-Extended, Meta-ExternalAgent, Bytespider, CCBot and cohere-ai. It also catches the failure a name-counting checker cannot see: a named AI-crawler group whose rules are stricter than your * group, so those agents are quietly fetching less than everyone else.
llms.txt: does yours exist, and is it valid?
An llms.txt is the curated summary an AI agent reads instead of crawling your whole site. The audit fetches yours and checks the shape the spec asks for: an # heading, a > blockquote summary, at least one ## section, and link lines in [Name](URL): Description form, and names which of those are missing rather than only telling you the file is there.
Open Graph and Twitter cards drawn, not just validated
Rather than only checking that the tags exist, the tool draws the card your page would actually produce on X, LinkedIn, Reddit, WhatsApp, and Discord, annotated with each platform's caching quirks.
Every fix named, and a prompt your AI coding agent can run
Every failed or warned check names its remedy: the exact tag to add, the exact header value to set. After a run, one button compiles the full list into a structured prompt for an AI coding agent, critical issues grouped first with a verification checklist at the end; another copies a plaintext report tagged line by line. Results get a shareable address (/seo/ followed by your URL) so one link re-runs the identical audit for anyone. The tool is free, and this site holds itself to it: our homepage description is budgeted to 160 characters so we pass our own check.