WebSecure

Recorded example · real WebSecure scanner · 19 Nov 2026 · test page made by HoopGram

What happens when you scan a site

A recorded scan of a HoopGram test page, step by step: what we looked at, what we found, and how you can check it yourself.

In one minute

What did you check?
One public page, scan.hoopgram.com/example/shop. 26 checks on what any browser receives: the connection and certificate, security headers, cookies, the page's own scripts and public DNS records. No login, no attack traffic.
What did you find?
Two real problems, both planted on purpose: an old jQuery with a bug that attackers are using, and a cookie sent without Secure. Plus one note for information.
How can I verify it?
Open the test page and both reports below. Every finding shows what we saw, and every software verdict links to its public source.
What if you keep watching?
Protect, $9 per site a month: a check every day, an email only when something new needs you, and an email when we confirm a fix.

The scan, step by step

  1. You enter an addressYou

    Uses: the address only. No account, no card.

    In this scan: https://scan.hoopgram.com/example/shop, entered at 05:38:46 UTC.

  2. We connectAutomatic

    Uses: DNS, the HTTPS response and the certificate. Private and internal addresses are refused, again on every redirect.

    In this scan: HTTPS with TLS 1.3, a valid Let's Encrypt certificate with 89 days left, and old TLS 1.0 and 1.1 refused.

  3. Public checksAutomatic

    Uses: only what any visitor's browser receives: response headers, cookies, the page and up to 12 of its own scripts, and public DNS records.

    In this scan: 26 checks, 11 requests, about one second. 22 passed, 3 found something, 1 could not run. Deep checks for exposed files and admin pages need a verified domain, so they show as "could not run", never as passed.

  4. Known-bug matchingAutomatic, only when a version is visible

    Uses: OSV.dev for known bugs in a package version, CISA's list of bugs attackers are using now, and endoflife.date for whether a version still gets security fixes.

    A version we could see:

    Seen in the page's scriptsjQuery 1.12.4
    OSV.dev4 known bugs
    One of themCVE-2020-11023
    CISA's listExploited, listed 23 Jan 2025

    A version we could not see:

    Server headernginx
    VersionHidden
    ResultNot matched

    When a version is hidden we say "not matched", never "safe". A match means this version contains the bug. It does not mean the site was hacked.

    Separate and optional: your package-lock.json. A URL scan only sees what the browser downloads, not the packages on your server. You can send the lockfile through the API or your coding agent, and we check every package the same way.

  5. The reportAutomatic

    Uses: the results above. An AI writes the explanations, but it cannot add findings. The checks are fixed code, so the same site gives the same result twice.

    In this scan: the verdict, two problems with evidence and fixes, the software list and every check, ready 19 seconds after the address was entered.

    Not safe: fix nowAn attacker could use the jQuery bug today. That is why one finding is enough for this verdict.

    Open the real report

One finding, before and after

The cookie problem is the easiest one to see for yourself. Without Secure, a browser can send the cookie over plain http://. Someone on the same public Wi-Fi could read it, and if it were a real login cookie, act as you. It is a medium problem, not a break-in on its own.

Before · problem found · medium

Scan s_kD3UuN5obbYI, 05:38 UTC

GET https://scan.hoopgram.com/example/shop
Set-Cookie: demo_session=…; Path=/example; HttpOnly; SameSite=Lax

After · same page, same check · passed

Rescan s_C2sP7hvJoCbB, 05:39 UTC

GET https://scan.hoopgram.com/example/shop
Set-Cookie: demo_session=…; Path=/example; HttpOnly; SameSite=Lax; Secure

The scanner hides cookie values (…) in every report. This one holds no data anyway.

The fix we deployed

The report's suggested fix: set Secure; HttpOnly; SameSite=Lax on session cookies. On the test page that was a one-word change, deployed at 05:39 UTC:

- Set-Cookie: demo_session=…; Path=/example; HttpOnly; SameSite=Lax
+ Set-Cookie: demo_session=…; Path=/example; HttpOnly; SameSite=Lax; Secure

The rescan of the same page then reported: Fixed (1) the cookie, Still open (2) jQuery and the mail note. We left jQuery old on purpose, so you can scan the page and see that finding yourself.

Check it yourself

When we cannot confirm a fix

Right after the fix, we scanned a different page on the same site, https://scan.hoopgram.com/. That page sets no cookie at all, so the cookie problem was simply not there. A careless comparison would call it fixed. Ours doesn't:

Could not verify (3)That scan checked a different page, so we cannot confirm whether the earlier findings were fixed. To confirm a fix, rescan the same page.

From scan s_TVcsBJsmfJap, 05:39 UTC. Only the rescan of the same page, above, could call the cookie fixed.

Where the free scan ends

The scan is done when the report is ready. Everything after that is your choice, and each choice is separate.

Fix it
You, your developer, or your coding agent. Each report has a "Copy fix prompt" button for Claude Code, Cursor or Codex.
Rescan
Free. Same page, same checks. Only then do we call something fixed.
Keep watching
Protect, $9 per site a month. A check every day, and the same day CISA adds a bug attackers are using. Your software list rechecked against OSV.dev, CISA and endoflife.date. An email only when something new needs you, and an email when we confirm a fix. A weekly deep scan once your domain is verified, with the verification kept fresh for you. Cancel any time. See plans

What no URL scan can see: pages behind a login, who can read whose data in your app, and code that never reaches the browser. Those need a different kind of review.

Scan your own site