Skip to content

Pre-launch audit with a prioritised fix list

Audit the whole site before launch and get a ranked fix list, then fix the blockers.

  • Free, no sign-up
  • 30 to 60 minutes with an agent
  • By Devvrat Hans
  • Version 1.0, updated Oct 26, 2026

1Your site and stack

Optional. The prompt adapts as you type; leave a field blank and the agent works it out.

2Copy it into your agent

pre-launch-audit.md75 lines, about 1,258 tokens

Paste into any coding agent: Claude Code, Cursor, Codex, Copilot, Windsurf or Cline.

# Pre-launch audit with a prioritised fix list

I am launching the production site soon. Audit this codebase the way a careful reviewer would on the day before a launch, rank every problem by how much it would hurt the launch, and then fix the blockers with me. Strangers will arrive from launch sites, social posts and search all at once: anything broken, slow, unclear or untrustworthy costs signups that do not come back.

Site: not given. Find the production URL in the project config, or ask me.

## Ground rules

1. Read before you write. Explore the parts of the codebase this job touches and tell me what is already in place.
2. Plan, then stop. Give me a numbered plan: what you will change, in which files, and why. Wait for my approval before editing anything.
3. Fix things at their source (the shared layout, the config, the generator), not with one-off patches page by page.
4. Keep the project's conventions: its framework APIs, file layout, naming and code style. Add no new dependency without asking.
5. Where something is already done well, say so and leave it alone.

## You will not

- Edit `.env` files, secrets or API keys, or print their values.
- Deploy, push, merge or publish anything.
- Delete data, or run migrations or scripts against a production database.
- Add tracking, third-party scripts or links to any site I did not ask for.
- Change copy, design or behaviour that this job does not need.
- Rewrite page copy without showing me the new wording first.

## Your stack

- Framework: Work out the framework, router and build output from the project files (package.json, lockfiles, framework config) before planning, and say which you found.
- Host: Work out where the site is hosted from the config files (vercel.json, netlify.toml, wrangler config, Dockerfile, fly.toml) and say which you found.

## Steps

### 1. Map the site

List every public route or page template, the shared layout(s), where metadata is set, and how the site is built and served. Note any pages that need sign-in. This map is what you audit against, so show it to me first.

### 2. Audit, without changing anything

Check each area below. For every problem, record the file, what is wrong, why it matters on launch day, and the fix.

- **Search and sharing:** a unique `<title>` and meta description on every public page; one canonical URL per page; Open Graph and X card tags with a 1200 x 630 image that exists; a favicon and apple-touch-icon; `robots.txt` and an XML sitemap that list the real public pages and no private ones.
- **AI assistants:** the main content is in the server-rendered HTML (not only after JavaScript runs); the homepage says in plain words what the product is and who it is for; there is an Organization or SoftwareApplication JSON-LD block.
- **Errors:** a real 404 page that returns status 404 (not 200); a 500 page; no broken internal links; no console errors on the main pages; forms that show a useful error when they fail.
- **Speed:** oversized images, images without width and height, render-blocking scripts in `<head>`, fonts loaded without `font-display`, and anything that will be slow under a burst of traffic.
- **Trust:** HTTPS everywhere with HTTP redirecting to it; a privacy policy and terms linked in the footer; a contact route; security headers (`Strict-Transport-Security`, `X-Content-Type-Options`, a `Content-Security-Policy` or a plan for one, `Referrer-Policy`).
- **Conversion:** the main call to action is visible without scrolling on a phone; sign-up works end to end; the pricing page, if there is one, matches what checkout charges.

### 3. Rank the findings

Sort everything into three groups and show me the list before fixing:

- **P0, blocks the launch:** broken flows, wrong or missing titles and share images on the pages people will share, a 404 page that returns 200, private pages in the sitemap, anything that loses data or leaks it.
- **P1, costs signups or rankings:** slow pages, missing structured data, unclear homepage copy, missing canonicals.
- **P2, polish:** everything else.

Give each finding a size (minutes, an hour, more) so I can choose.

### 4. Fix the P0s, then the P1s I approve

Fix at the source: a missing description belongs in the shared metadata helper with a per-page value, not pasted into each page. After each fix, re-check that item and tick it off in the list. Stop and ask before any fix that changes copy the visitor reads.

## Done means

- [ ] The audit list exists, ranked P0, P1 and P2, with a file and a fix for every item.
- [ ] Every P0 is fixed or explicitly deferred by me.
- [ ] Every public page has a unique title, a description, a canonical URL and working share tags.
- [ ] An unknown URL returns a 404 status with a helpful page.
- [ ] The build and existing tests still pass.

## Report back

When you finish, give me:

- what you changed, file by file;
- anything you skipped, and why;
- anything I must do by hand (DNS records, dashboards, accounts, deploys);
- how to check each change on the live site.

One-line version

For agents that can read a URL. It fetches this prompt with your settings applied.

Read https://abacklaunch.com/prompts/pre-launch-audit.md and follow it for this project.

3Check the result

When the changes are live, run these free checks. Add your site URL above and each one opens already running on it.

This prompt will not

  • Edit .env files, secrets or API keys, or print their values.
  • Deploy, push, merge or publish anything.
  • Delete data, or run migrations or scripts against a production database.
  • Add tracking, third-party scripts or links to any site I did not ask for.
  • Change copy, design or behaviour that this job does not need.
  • Rewrite page copy without showing me the new wording first.

It reads your code first, shows you a plan, and waits for your approval before it changes anything.

What you get

  • A ranked audit of your codebase: P0 blockers, P1 issues and P2 polish, each with a file and a fix
  • The blockers fixed at their source, not page by page
  • A list of anything you must do by hand, like DNS or dashboard settings

How to use this prompt

  1. 1

    Describe your stack

    Add your site, framework and host. The prompt adapts; blank fields are left for the agent to work out.

  2. 2

    Paste it into your agent

    Open your project in Claude Code, Cursor, Codex, Copilot, Windsurf or Cline, and paste the prompt or the one-line version.

  3. 3

    Approve the plan

    The agent reads your code, shows you a plan with the files it will touch, and waits. Change anything you disagree with.

  4. 4

    Check the result

    When the changes are live, run the free checks below on your site to confirm each fix.

What the agent does

I am launching the production site soon. Audit this codebase the way a careful reviewer would on the day before a launch, rank every problem by how much it would hurt the launch, and then fix the blockers with me. Strangers will arrive from launch sites, social posts and search all at once: anything broken, slow, unclear or untrustworthy costs signups that do not come back.

  1. 1. Map the site

    List every public route or page template, the shared layout(s), where metadata is set, and how the site is built and served. Note any pages that need sign-in. This map is what you audit against, so show it to me first.

  2. 2. Audit, without changing anything

    Check each area below. For every problem, record the file, what is wrong, why it matters on launch day, and the fix.

    • Search and sharing: a unique <title> and meta description on every public page; one canonical URL per page; Open Graph and X card tags with a 1200 x 630 image that exists; a favicon and apple-touch-icon; robots.txt and an XML sitemap that list the real public pages and no private ones.
    • AI assistants: the main content is in the server-rendered HTML (not only after JavaScript runs); the homepage says in plain words what the product is and who it is for; there is an Organization or SoftwareApplication JSON-LD block.
    • Errors: a real 404 page that returns status 404 (not 200); a 500 page; no broken internal links; no console errors on the main pages; forms that show a useful error when they fail.
    • Speed: oversized images, images without width and height, render-blocking scripts in <head>, fonts loaded without font-display, and anything that will be slow under a burst of traffic.
    • Trust: HTTPS everywhere with HTTP redirecting to it; a privacy policy and terms linked in the footer; a contact route; security headers (Strict-Transport-Security, X-Content-Type-Options, a Content-Security-Policy or a plan for one, Referrer-Policy).
    • Conversion: the main call to action is visible without scrolling on a phone; sign-up works end to end; the pricing page, if there is one, matches what checkout charges.
  3. 3. Rank the findings

    Sort everything into three groups and show me the list before fixing:

    • P0, blocks the launch: broken flows, wrong or missing titles and share images on the pages people will share, a 404 page that returns 200, private pages in the sitemap, anything that loses data or leaks it.
    • P1, costs signups or rankings: slow pages, missing structured data, unclear homepage copy, missing canonicals.
    • P2, polish: everything else.

    Give each finding a size (minutes, an hour, more) so I can choose.

  4. 4. Fix the P0s, then the P1s I approve

    Fix at the source: a missing description belongs in the shared metadata helper with a per-page value, not pasted into each page. After each fix, re-check that item and tick it off in the list. Stop and ask before any fix that changes copy the visitor reads.

Done means

  • The audit list exists, ranked P0, P1 and P2, with a file and a fix for every item.
  • Every P0 is fixed or explicitly deferred by me.
  • Every public page has a unique title, a description, a canonical URL and working share tags.
  • An unknown URL returns a 404 status with a helpful page.
  • The build and existing tests still pass.

Questions

How is this different from running an SEO checker?

A checker reads one page from the outside. This prompt reads your codebase, so it finds the cause (a layout with no description, a sitemap built from the wrong list) and fixes it once for every page. Run a checker afterwards to confirm the live result.

Will the agent change my site without asking?

No. The prompt tells the agent to map the site, audit without changing anything, show you a ranked list, and wait for your approval before each fix. It also lists things it must never do, like editing secrets or deploying.

How long before launch should I run it?

A week before is ideal: enough time to fix the P0s and P1s calmly. Run it again the day before to catch anything new.

Does it work with no-code builders?

Pick Webflow, Framer or WordPress as your framework and the page gives you the same audit as a checklist you can follow in the builder's settings.

Does this prompt add links to Aback Launch to my site?

No. Nothing in the prompt mentions Aback Launch or asks your agent to link to us. The only exception is an optional badge step, which stays off unless you tick it. Some prompt libraries slip their own link into your codebase by default; we think that should always be your choice.

Which coding agents does it work with?

Claude Code, Cursor, Codex, GitHub Copilot, Windsurf, Cline, and any agent that can read your project and edit files. You can copy it as a plain prompt, save it as a Claude Code skill or a Cursor rule, or add it to AGENTS.md.

Is it safe to run on my real codebase?

The prompt makes the agent read first, show a plan and wait for your approval, and it forbids touching secrets, deploying and deleting data. Still review the changes and run your tests before you deploy, as with any change.

Changelog

  • v1.0First version.

Launch your startup on Aback Launch

Get a launch page that search engines and AI assistants can read, a place in the weekly launch list, and a review by a real person. Free to join.