Skip to content

Website QA / ChatGPT

Audit your website with ChatGPT.

Give ChatGPT a way to inspect your actual website. Afterward checks public pages for broken links, missing metadata, configuration issues, and other launch-readiness problems, then returns structured findings your assistant can help you understand.

  • Free one-off audits
  • No Afterward account needed for one-off scans
  • Every finding shows what was observed
  • Nothing to install on your website
RESPONSEPAGEHEADFIG. 0101BROKEN LINKlinks.broken-internal02MISSING ALTimages.alt-text03MISSING CANONICALseo.canonical-present04SECURITY HEADERsecurity.csp01020304
Fig. 01 — Example inspection. Illustrative findings, not a scan of a real site.
Your assistant calls scan_website with a public URL. Afterward inspects the pages and returns findings with evidence. Illustration.

01 / How it works

A website audit. Inside your conversation.

  1. 01 / Connect

    Connect Afterward.

    In ChatGPT, open Plugins and look for Afterward. If it isn’t listed for your account, you can add its public server instead. Our connection guide covers both.

  2. 02 / Scan

    Ask ChatGPT to check your site.

    Give it a public URL and ask for a pre-launch audit. Type @ and choose Afterward to make sure it’s used.

  3. 03 / Review

    Review the findings.

    Afterward inspects the site and returns a launch-readiness score and findings with evidence. ChatGPT helps you understand them and decide what to fix first.

Example prompt

Use Afterward to check https://example.com before I launch it. Identify technical problems and tell me what I should fix first.

Who does what: Afterward performs the website checks, the same way every time. ChatGPT helps you interpret the results. ChatGPT doesn’t scan your site itself.

02 / What Afterward checks

More than a second opinion. Actual website checks.

Afterward runs 50+ deterministic checks against your site’s real responses. Each finding has a stable check ID and the evidence behind it.

  1. Technical configuration

    Reachability, HTTPS, redirect chains, response status codes, the mobile viewport and HTML basics.

    technical.https · technical.redirect-chain
  2. SEO & discovery

    Page titles, meta descriptions, canonical URLs, noindex directives, Open Graph tags, robots.txt and the XML sitemap.

    seo.canonical-present · discovery.sitemap
  3. Security headers

    Common hardening headers, including HSTS and Content-Security-Policy, plus mixed content and cookie attributes.

    security.hsts · security.csp
  4. Links & page quality

    Broken internal links, a sample of external links, development URLs, placeholder text and malformed form actions.

    links.broken-internal · content.placeholder-text
  5. Accessibility basics

    Missing image alt text, links without accessible names, unlabeled form fields and the declared page language.

    images.alt-text · forms.field-labels
  6. Analytics detection

    Common analytics tags in the returned HTML, retired Universal Analytics tags and uneven coverage across pages.

    analytics.detected

Read the full scan scope

03 / An example result

Findings you can actually work with.

Warninglinks.broken-internalIllustrative example, not a real scan

Broken internal link

Affected URL
https://example.com/pricing
Observed response
HTTP 404
Severity
Warning
What it means
An internal link points to a page that doesn’t exist.
Suggested next step
Correct the link’s destination or restore the missing page, then run another audit.

04 / What the scan doesn’t do

A clear scope. No false confidence.

  • Server-delivered HTML onlyAfterward reads the HTML your server returns: 10 pages by default, up to 25, at most two links from the homepage.
  • No JavaScriptClient-side JavaScript isn’t executed, so content that only appears after scripts run may be missed.
  • Public pages onlyIt doesn’t sign in, so pages behind a login aren’t checked.
  • No form submissionsForms are inspected, never submitted.
  • Not a penetration testIt checks common configuration and hardening issues. It doesn’t probe for vulnerabilities.
  • Not exhaustiveA clean result doesn’t mean every problem was found.
  • Not an assessmentThe launch-readiness score is an indicator, not a full accessibility, security, SEO or compliance assessment.

05 / Why connect Afterward

Give your assistant evidence, not guesses.

Asked to critique a website, ChatGPT works from what it’s shown and what it knows about good practice. With Afterward connected, it also gets what your server actually returned, checked the same way every time. Afterward doesn’t replace ChatGPT; it gives it concrete evidence for a specific job.

  • Actual HTTP observations

    Status codes, redirects, headers and the HTML your server returned, not a description of them.

  • Consistent checks

    The same responses always produce the same findings and the same score.

  • Structured findings

    Each result has a stable check ID, a severity and the affected URLs.

  • Inspectable evidence

    Findings include what Afterward observed, so you can verify them yourself.

  • A clear scope

    The limits above are part of every result, not small print.

  • Help in plain language

    ChatGPT explains the findings and helps you decide what to fix first.

Before you connect it. Straight answers.

What Afterward does in ChatGPT, and what it deliberately doesn’t.

Can ChatGPT audit a website?

On its own, ChatGPT can review what you paste or describe and suggest general best practices. With Afterward connected, ChatGPT can ask Afterward to inspect a public website: Afterward fetches the pages, runs its checks and returns findings with evidence, and ChatGPT helps you understand them.

How do I use Afterward in ChatGPT?

In ChatGPT, open Plugins and search for Afterward. If it’s listed for your account, install it. If it isn’t, you can add Afterward’s public server as a custom MCP server where your account allows it. Then start a new chat, type @, choose Afterward and ask it to check a public URL. Our connection guide has both routes step by step.

Is Afterward free?

One-off audits are free and need no Afterward account. They’re rate limited, currently to 5 every 10 minutes. Ongoing monitoring is a paid plan: Pro is $19 a month and Agency is $79 a month.

Does Afterward need access to my website’s code?

No. Afterward fetches your public pages over the internet, as any visitor can. There’s nothing to install on your site and no code to share.

Can Afterward check a website that requires a login?

No. Afterward only checks public pages. It doesn’t sign in, so anything behind a login isn’t inspected.

Does Afterward perform penetration testing?

No. Afterward checks common configuration and security-hardening issues, such as missing headers. It doesn’t probe for vulnerabilities and doesn’t guarantee that a site is secure.

Can I use Afterward without ChatGPT?

Yes. Afterward works with Claude, Codex and other MCP-compatible assistants, using the same public server.

What’s the difference between an audit and ongoing monitoring?

An audit is a one-off check of your site as it is now. Monitoring, on the Pro and Agency plans, saves a baseline, checks your site on a schedule, compares each scan with the last and emails you when something new breaks.

Your next launch deserves a second set of checks.

Connect Afterward to ChatGPT, provide a public URL, and get a clearer picture of what needs attention.

Every scan covers
  • HTTPS
  • robots.txt
  • Sitemap
  • Canonicals
  • Open Graph
  • HSTS