Extension

# Privacy policy

This covers the CrawlCheck browser extension. Every claim below is checkable by reading the source, and each one says which version it applies to.

## What reaches crawlcheck.io, from version 1.0.0

- **The domain you ask about, when you press “Two vantages”.** The extension sends the site’s domain to `GET /api/flow` to fetch CrawlCheck’s edge reading of it, then compares that reading with what your own connection receives. Only on that button; never in the background. No page content, no path, no cookies.

- **Anonymous usage counts**, to `POST /api/ext/event`: the event name (installed, updated, popup opened, a tool pressed, a verdict’s class counts), the extension version, a random install id created on your device, and small counters. **Never a URL, a domain, page content, a licence key or anything typed.** The server refuses any field that could carry one, and stores daily totals only. Switch it off with the checkbox in the popup footer; the choice is honoured at once.

- **Your licence key**, if you enter one: `GET /api/licence/check`, at most once every 24 hours, with the key and nothing else.

- **A Data Axle listing page, in versions 1.0.1 to 1.0.7 only** (removed in 1.0.8, 11 October 2026; 1.0.8 and later make no request to Data Axle), only when you pressed “Data Axle listing” *and* had saved a licence key. What those versions did: the extension reads the phone number the current page declares, fetches Data Axle’s public search page for that number from your browser, and posts that search page and the site’s domain to `POST /api/nap/observe`, so the site’s CrawlCheck report can show whether the business is listed. The server accepts it only for a domain on your licence. Without a saved key the answer is shown in the popup and nothing is sent.

- **A listing form fill, in version 1.0.7 and from version 1.0.9** (1.0.8 does not have it), only when you press “Fill listing form” with a saved licence key. The values already in the form are read only to compare them with your record; they are shown to you in the popup and never sent. the extension asks `GET /api/v1/business/submit-kit` for your business’s own record (name, phone, website, address or service area, from your site and linked Business Profile) and types it into the add or claim form open in your tab. Nothing is read from that page and sent anywhere, the form is never submitted by the extension, and contact names, email, passwords and captchas are never touched. Only if you then press “I submitted it” does the extension send `POST /api/v1/business/submissions`: your domain, which directory, that page’s hostname and the names of the fields it filled (never their values), so CrawlCheck can check the directory at its next daily reading.

Versions up to 0.9.1 sent none of the first two: they made only the licence check.

## The intelligence card (Pro), from version 1.0.19

When the popup opens on a site, **the hostname of the site in your active tab** goes to `GET /api/extension/intel` together with your licence key in a request header (never in a URL), so the card can show what CrawlCheck’s intelligence engine holds on that site *for your licence*: for the site you registered as yours, its own changes, plugin exploit alerts, third-party script changes, cloaking and lookalike domains; for a competitor you track, its public prices, service areas, reviews and recent changes. Without a Pro licence the same call returns only the offer. Pressing **Watch this as my site** or **Track as competitor** sends that hostname to `POST /api/extension/intel/own` or `/competitor` and stores it against your licence (up to the competitor limit of your plan); **Stop tracking** removes it. The engine then reads those sites’ public pages on its own schedule. The hostname only: never the path, the page, a cookie or anything typed.

## Home vantage, from version 1.0.20

A switch in the popup footer, **off until you turn it on**, and it works only while the Two-vantages share switch is on too. While it is on, twice a day the extension runs **Two vantages** by itself on the site you marked as yours and the sites on your watch list (at most ten, one at a time, nothing else), from your own connection, and files each reading exactly as a pressed run does: to `POST /api/vantage/observe`, signed with this install’s key, with the same fields (hostname, the page address measured, and per identity the status, bytes, word count, body hash and verdict). Never page text, cookies, anything typed or your licence key. The purpose: CrawlCheck’s own readings come from a data centre, Tor and AWS; this adds a home connection for your own sites. The popup shows when each site was last read. Switch it off and the schedule is removed at once.

## The AI view slider, Pro checkout and theme, from version 1.0.23

**The AI view slider**: when you press it (or Alt+Shift+S), your browser fetches the page in your active tab once more, without cookies, straight from that site, and compares the answer with what is on your screen. That request goes to the site you are on, not to CrawlCheck. The word counts it produces stay in your browser’s session memory and are gone when the browser closes. **Pro**: pressing **Pro** opens a Stripe checkout page. The extension makes a random reference on first use, keeps it in your browser and adds it to the checkout address, so the payment can be matched to this install without a key to type. It then sends the same reference to `GET /api/extension/pro` to learn whether Pro is on, and receives the licence key that belongs to the payment. The reference is random: it says nothing about you or the pages you visit. Card details are typed on Stripe’s page and never pass through the extension. **Restore Pro** sends the email address you type and that reference to `POST /api/extension/pro/restore`; if the address holds Pro, a confirmation link is mailed to it, and nothing else is done with the address. **Theme**: your light or dark choice is stored in your browser only. **Grades for sites not yet scanned**: when the extension asks for the grade of a site with no scan on record, the hostname alone is noted for up to three days so the site can be scanned and its grade shown next time. Nothing ties that hostname to you or your browser. **Switches, from version 1.0.23**: grades as you browse, chips on search results and AI answers, the daily page reading, shared readings and anonymous usage counts are on unless you switch them off. The popup says so the first time it opens after an install or update, with a button to Settings, where each switch is honoured at once. A switch you had already turned off stays off. Home vantage stays off until you turn it on.

## The agent view and the registry, from version 1.0.8

**The domain of the site in your active tab**, to `GET /api/v1/resolve`, only when you open the **Agent view** tab, so it can show what CrawlCheck tells an AI agent about that site. The domain only: never the path, the page, a cookie or anything typed. The extension also reads CrawlCheck’s public key directory (`/.well-known/http-message-signatures-directory`) to check the answer’s signature in your browser. If you press a draft button there, the domain goes to `/api/tool/llms`, `/api/fix/robots` or `/api/fix/entitymap` to draft that file. **Join the registry** opens `crawlcheck.io/registry/<domain>/claim` in a new tab; the extension itself sends nothing.

## The registry record and the globe, from version 1.0.22

**The hostname of the site in your active tab**, to `GET /api/registry`, when you open the **Registry** view in the popup, so it can show that site’s public registry record. The request is made only while the **Grades as you browse** switch is on; with the switch off it is not made. Hostname only: never the path, the page, a query string, a cookie or anything typed, and the request is sent without cookies. **The globe**: when you open the **Globe** view, the extension fetches `GET /api/globe.json`, the same public file the globe on crawlcheck.io draws from. That request names no site and is sent without cookies. The answer is kept on your device for 30 minutes so the small globe in the popup can be drawn without another request.

## Grades as you browse, from version 1.0.3

**The hostname of the page in your active tab**, to `GET /api/extension/site`, so the toolbar icon can show that site’s grade if it has ever been measured. Hostname only — never the path, the page, a query string, a cookie or anything typed. The answer is the same public record a report shows. Versions through 1.0.20 shipped with it on by default, stated on first run. In versions 1.0.21 and 1.0.22 it was off until switched on. From version 1.0.23 it is on again unless you switch it off, and the popup says so the first time it opens. The switch is in the popup (the footer through 1.0.21, Settings from 1.0.22) and is honoured at once. Sites you add to your watch list are looked up the same way, in one request of up to 25 hostnames, when you open the popup. The server does not store which hostnames were asked about.

## Daily page readings from sites you visit, from version 1.0.17

Once a day per site, about six seconds after a page in your active tab finishes loading, the extension takes the same page reading a check shares (from version 1.0.11): field vitals, request and third-party counts, script errors, shadow, hidden and noscript word counts, controls an agent cannot operate, image counts and the share of text that only appears after JavaScript. It is signed with the install’s key and sent to `POST /api/vantage/page`. Numbers and the hostname only, plus the page address you were on with its query string removed — never page text, a cookie or anything typed. To measure the JavaScript-only share it fetches the page once more with no cookies and your browser’s own user-agent. Never in incognito windows, at most 40 sites a day, and not while the **readings**, share or grade switch in the popup footer is off. A clean Chrome profile does not run the extension in incognito. If you allow it there, this daily reading still does not run; the grade and chips switches still send hostnames. Every install is told once, with the off switch beside the notice.

## Chips on search results and AI answers, from version 1.0.16

On Google, Bing and DuckDuckGo results pages and on ChatGPT, Perplexity, Claude, Gemini and Copilot, the extension reads the **hostnames of the outbound links** on the page (the results, or the sources an answer cites) and asks `GET /api/extension/sites` for their grades, up to 25 hostnames per request, the same lookup the toolbar icon makes. Hostnames only — never your search words, your prompt, the answer, a path, a cookie or anything typed. Nothing is requested from the listed sites themselves. The server does not store which hostnames were asked about. It is on while both the **chips** switch and the grade switch in the popup footer are on, and either switch turns it off at once.

On first install the extension opens a welcome page that asks which site is yours; what you type is kept in your browser as your site and looked up the same way.

## AI Overviews you see, from version 1.0.9

On a Google results page, the extension reads the AI Overview Google already showed you: whether there was one and which sites it links to. It makes no request of its own and reads nothing else on the page. The search words, the cited links and the time are kept in your browser (the last 300 searches) so the **AI Overviews** tab can show whether your site is cited; **Clear this list** deletes them. Nothing leaves your device unless you tick **Share** in that tab (off by default). With it on, the search words and the cited links go to `POST /api/aio/observe` to build the AI Overview citation index (`/api/aio?domain=`). The words are lowercased and stored without your address, licence key or any identifier; searches that look like they contain personal details (an email address, a phone number, long numbers, “my”/“me”) are dropped in the extension and again on the server.

## What stays on your device

## Page readings, from version 1.0.11

When you press **Check this page** with sharing switched on, a reading of that page as your browser saw it goes to `POST /api/vantage/page`: the hostname and page address, field Web Vitals from your visit (LCP, INP, CLS, TBT, TTFB), counts from the network timeline (requests, failures, render-blocking, third-party hosts and the five heaviest third-party hostnames), counts of script errors, promise rejections and Content-Security-Policy violations, word counts for text inside shadow roots, frames and hidden elements, counts of links, forms, buttons and images an agent could not use, and whether a service worker or a consent overlay was present. Counts and hostnames only: never the page text, the error messages, cookies, anything typed or your licence key. From version 1.0.11 the extension also keeps, in the tab's memory only, a count of script errors and policy violations as each page runs, so that this reading can include them; those counts leave the tab only inside a reading you send. The same per-install key signs it and the same switch controls it.

## Signed Two-vantages readings, from version 1.0.4

When you press **Two vantages** with sharing switched on, the reading from your connection goes to `POST /api/vantage/observe`: the hostname, the page address that was measured, and for each identity the HTTP status, byte count, visible word count, a hash of the body and the verdict class. Never the page content, cookies, anything typed or your licence key. Your connection is recorded as country and network name only; the address is never stored. Each install signs its readings with its own key, made on first use and kept in your browser, so a reading can be shown to come unaltered from one install; the key does not identify you. The site’s report then shows your network beside CrawlCheck’s. Sharing is on by default and the switch to turn it off is in the popup footer.

- Measurements you run, held in the browser’s local and session storage. Apart from a shared Two-vantages reading (above), they are never sent anywhere, and you can clear them by removing the extension.

- Your licence key, if you have one, the random install id, and your usage-count choice.

- The last run, so the popup reopens where you left it.

## Requests to sites you measure

When you measure a page, the extension requests it as each named crawler and as your browser. Those requests are sent with `credentials: "omit"`, so **your cookies, sessions and logins are never attached**. The extension identifies itself honestly and does not impersonate an operator’s crawler to obtain access it would not otherwise be given.

## Sharing and sale

Nothing is shared with, sold or transferred to third parties; the usage counts stay at crawlcheck.io. Nothing is used for advertising, creditworthiness or lending, or to train anything. There is no remote code and no advertising network.

## Permissions, and the feature each one exists for

- `declarativeNetRequest` — a request cannot set its own `User-Agent` from a page, so a session rule is the only supported way to ask for a page as a named crawler. One rule exists at a time and it is removed in a `finally` block.

- `scripting` and `activeTab` — read timings and the rendered DOM of the tab you are looking at, when you ask. It reads and returns, and it never submits a form. Two features change what you see in the tab when you ask for them: AI X-ray draws outlines over the page, and Fill listing form fills a directory’s form fields for you to check and submit yourself.

- `tabs` — read the active tab’s address so the popup measures the site you are on, and clear a tab’s saved check when it navigates away or closes.

- `storage` — the local items listed above.

- `contextMenus` — one right-click item.

- `alarms` — re-read the watch list every six hours, so the toolbar dot shows when a watched site’s grade moves.

- Host access to all sites — you choose which site to measure and it is not a site we can predict. That choice is the entire function.

## Support

For help with the extension, a bug report or a question, email [hello@crawlcheck.io](mailto:hello@crawlcheck.io). What each tool measures and how to use it is on [the extension page](/extension).

## Changes and contact

Material changes are published on this page with the date they took effect. Questions: [hello@crawlcheck.io](mailto:hello@crawlcheck.io).

Last updated 7 November 2026 (1.0.20: AI X-ray, which runs in the tab and sends nothing; 1.0.19: the Intelligence card, which sends the hostname and your key header when the popup opens and only if you saved a Pro key; 1.0.18: no new request; a slow paint is not put on the site when other extensions are running; 1.0.17: daily page readings from sites you visit; 1.0.16: chips on search results and AI answers, the welcome page; 1.0.9: Fill listing form, as in 1.0.7; 1.0.8: the agent view and Join the registry, Data Axle check removed; 1.0.1: the Data Axle listing check; 1.0.0: Two vantages and anonymous usage counts). The extension currently published is version 1.0.23.

## Keep reading

[What the extension measures The second vantage point, and why one scan cannot settle a refusal.](/extension)[What the server side collects A separate question from the extension, answered separately.](/policy)[How measurements are verified Nothing here depends on trusting our word.](/proof)

We read your site the way crawlers do, from our own network, and every report says exactly what was asked for and what came back. [Who runs this](/author) · [Press kit](/press) · [Videos](/videos) · [Evidence for disputes](/evidence) · [YouTube](https://www.youtube.com/@crawlcheck) · [hello@crawlcheck.io](mailto:hello@crawlcheck.io)

#### How we measure, and our own machine files

CrawlCheck reads every site as itself. Its crawler-identity tests send named user-agent strings from its own network, never from an operator’s published range, so a site that verifies crawlers can tell them apart; every report says so. Results reflect what an anonymous non-browser client receives. [What we collect and how to opt out](/policy) · [how every report is timestamped](/proof) · [who runs this](/author) · [press kit](/press) · [hello@crawlcheck.io](mailto:hello@crawlcheck.io).

**Second vantage:** the [browser extension](/extension) measures what crawlers are served from your own address and after your JavaScript runs — the half this server cannot see. From 1.0.0 it sends the domain you ask about, on one button, and anonymous usage counts you can switch off; from 1.0.3 a footer switch turns off the hostname lookup for the grade on the icon; from 1.0.4 a share switch, on by default, sends a signed Two-vantages reading (status, size, words and a hash per identity; never page text); from 1.0.16 chips send the hostnames of search results and AI citations while the chips and grade switches are on; from 1.0.17 a readings switch, on by default, sends one signed page reading a day (numbers, the hostname and the page address with the query removed; never in incognito; at most 40 a day) and only while the share and grade switches are also on. Earlier versions send nothing.

**Our own machine layer for the agentic web:** [llms.txt](/llms.txt) · [llms-full.txt](/llms-full.txt) · [agents.md](/agents.md) · [SKILL.md](/SKILL.md) · [entitymap.json](/entitymap.json) · [media kit](/.well-known/media-kit.json) · [MCP server card](/.well-known/mcp/server-card.json) · [API catalog](/.well-known/api-catalog) · [OpenAPI](/openapi.json) · [robots.txt](/robots.txt) · [sitemap.xml](/sitemap.xml) · [security.txt](/.well-known/security.txt) — every one named in robots.txt, which is exactly what we check for on yours.

**Pages:** [/](/) · [Listing check](/listing-check) · [SDK](/sdk) · [Certificates](/certificates) | **Blog:** [One in four MCP hosts changed its tools in four days. Nothing told the agent.](/blog/one-in-four-mcp-hosts-changed-its-tools-in-four-days) | **Glossary:** [AI classifiers](/glossary/ai-classifiers) · [tdmrep.json](/glossary/tdmrep-json) · [Image sitemap](/glossary/image-sitemap) · [Semantic HTML](/glossary/semantic-html) · [Manifest store](/glossary/manifest-store) · [DataCatalog](/glossary/datacatalog) | **Directory:** [Wandb](/directory/wandb)

[Contact](/contact) · [Partnerships](/partnerships) · [Terms of service](/terms) · [Refund policy](/refunds) · [Trust center](/trust) · [Subprocessors](/subprocessors) · [Changelog](/changelog) · [Report an error](/report-error)
