> ## Documentation Index
> Fetch the complete documentation index at: https://docs.usecliro.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Site Check
> Audit your website for the artifacts and signals that make it readable and citable by AI models.
The **Site Check** verifies that your site has the Cliro-recommended artifacts published and is technically ready for AI discovery. Unlike a traditional SEO audit aimed at search engines, Site Check focuses on the signals AI assistants (ChatGPT, Gemini, Claude, Perplexity, Grok) rely on to read, understand and cite your content.
Run it after publishing your files to confirm everything is in place — and again whenever you change your site, so you catch anything that drifted out of sync.
## When to run a check
You just generated and deployed your `llms.txt`, `robots.txt`, `sitemap.xml` or the Ask AI script and want to confirm they're live.
You redesigned, migrated, or changed your CMS — re-run to make sure the artifacts and meta tags survived.
A monthly check catches regressions, like a deploy that overwrote your `robots.txt`.
Confirm your technical foundation is solid before you invest in prompts and content.
## Run a check
Go to **Actions → Site Check**. The page shows the date of your last run and the result of each item from that run.
Click **Start Site Check**. The audit fetches your site and inspects each signal. It takes up to \~30 seconds.
Each item resolves to a status (see [Reading the results](#reading-the-results)). Your overall score reflects how many checks passed.
Site Check reads your **live, public** site at your confirmed domain. If your site is behind authentication, staging-only, or blocks crawlers entirely, the checks can't fetch it and will report as not found or error.
## What it checks
Each card below is one signal in the audit. Together they answer a single question: *can an AI model find, fetch and trust your content?*
A published `llms.txt` gives AIs a curated map of your most important pages and how to interpret them. Generate or update it from [llms.txt](/en/actions/llms-txt).
Confirms your crawl rules explicitly allow the AI agents you want to reach you. A missing or over-restrictive file silently blocks AI crawlers. Manage it from [robots.txt](/en/actions/robots-txt).
A served sitemap lets crawlers discover every page instead of only what they stumble into via links. Build one from [Sitemap](/en/actions/sitemap).
Detects whether the conversational [Ask AI](/en/actions/ask-ai) widget is embedded on your pages, so visitors can query your site directly.
Your homepage content stays under 100 KB so models can analyze the whole page. Bloated pages get truncated, and your key message may fall outside the window.
Verifies your site has a title and meta description. These are the first things a model reads to understand what a page is about.
Compares the content visible without JS to the JS-rendered version. Many AI crawlers don't execute JavaScript, so content that only appears after JS runs is invisible to them.
A content-level read of your pages to gauge how clearly your topics, entities and value proposition come through to a model.
## Reading the results
Each item shows a status:
| Status | Meaning | What to do |
| ------------------------------------------ | -------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------ |
| **Found and matches Cliro** | The artifact is published and matches your confirmed content. | Nothing — this item is healthy. |
| **Found but does not match Cliro** | Published, but the content differs from what Cliro recommends/confirmed. | Re-publish the version from the matching tool, or confirm the live version in Cliro if the change was intentional. |
| **Found (no official content to compare)** | Published, with nothing to compare against. | Generate and confirm the recommended content in the matching tool so future checks can validate it. |
| **Not found** | The artifact isn't published. | Create and publish it via the matching tool in [Actions](/en/actions/overview). |
| **Not checked / Error** | The check hasn't run, or an error occurred (site unreachable, timeout, blocked). | Re-run the check. If it keeps failing, confirm the site is publicly reachable. |
Treat a low score as a checklist, not a grade. Fix each "Not found" or "does not match" item via the matching tool, then re-run the check. The score is meant to reach and stay at the top — not to be a one-time snapshot.
## Fixing each item
The file isn't being served at the expected path. Generate it in the matching tool ([llms.txt](/en/actions/llms-txt), [Sitemap](/en/actions/sitemap)), publish it to your site's root, and re-run. If you use a CMS, make sure the file is deployed and not blocked by a redirect rule.
Your `robots.txt` exists but disallows the crawlers you want. Open [robots.txt](/en/actions/robots-txt) to review which AI agents are allowed and adjust the rules.
Your page renders its main content client-side. Consider server-side rendering or pre-rendering for crawlers, or ensure the key content (headings, copy, links) is present in the initial HTML.
Your homepage HTML is large enough that models may truncate it. Trim unused markup, defer non-essential scripts, and keep your core message high in the page.
The widget snippet isn't present in your page's HTML. Copy the embed from [Ask AI](/en/actions/ask-ai) and paste it before the closing `` tag, then re-run.
## Frequently asked questions