# Front-Door Review Stack: catch machine-made tells, UX breaks and invented claims before a page goes live

> A review gate for agent-built landing, pricing, product and profile pages: a hard floor that blocks publishing, a truth pass for every claim, 20 UX principles as yes or no questions, 36 machine-made tells with fixes, a real browser pass and a fresh-eyes reviewer.

- URL: https://paperchasewebb.ai/bots/marketplace/items/front-door-review-stack-page-review-before-publish/
- Type: Prompt pack
- Category: Vibe coding
- By: PaperChaseWebb
- Works with: Claude Code, Codex, Cursor (named in the listing); Plain .md
- Price: $39 USD
- Difficulty: Intermediate
- Setup time: 30 min
- Time to result: First full review of a live page in one session
- Format: Markdown (.md) file, front-door-review-stack-page-review-before-publish.md
- Version: 1.0, updated 2026-10-27
- License: Personal and commercial use on your own and your clients' pages. Do not resell or republish this pack.

## What you end up with

No page goes live with a broken form, an invented testimonial or a pile of template defaults, and every tell you keep is a logged choice.

## What you get

- 14 sections in one .md file, about 16 pages
- A five-pass review order: hard floor, truth, usability, tells, polish
- Nine hard floor checks that block publishing
- An EVIDENCE.md template for every claim, quote, logo and number
- 20 UX principles turned into review questions
- 36 machine-made tells grouped by color, type, layout, parts, copy, trust and motion, each with a fix
- A Playwright browser pass at three widths with forms, links and an axe scan
- A fresh-eyes reviewer setup, REVIEW.md and an overrule log
- Profile page checks, a rules block, a page-reviewer SKILL.md and 12 prompts
- Linked credits for the 8 public sources we studied

## Who it is for

Builders and agencies whose coding agent makes pages fast and who want a fixed review gate so nothing broken, false or generic-looking goes live.

## Tools needed

- Claude Code, Codex or Cursor: builds the browser pass and fixes what the review finds
- Playwright: real browser screenshots, console, links and forms
- axe-core: automated accessibility scan
- You need: A page your agent built, local or staged
- You need: Node.js for Playwright
- You need: An agent that can run commands

Cost to run: Free. Playwright and axe-core are open source. Uses the agent plan you already pay for.

## Sample prompt: Count the tells (Claude Code, Codex, Cursor)

```
Look only at these screenshots and the page text. Go through the 36 machine-made tells and list which are present and where.
Skip any listed in OVERRULES.md.
```

Screenshots, not code. Judge what a visitor sees.

## Sample prompt: Fresh-eyes review (Claude Code, Codex, Cursor)

```
You have not seen this page being built. Using the page-reviewer skill, review these screenshots, page text and EVIDENCE.md and write REVIEW.md.
```

A builder grading its own page sees what it meant to build.

## Free preview of front-door-review-stack-page-review-before-publish.md

````markdown
# Front-Door Review Stack: catch machine-made tells, UX breaks and invented claims before a page goes live

A PaperChaseWebb pack, version 1.0. Written for Claude Code, Codex and Cursor, and plain Markdown, so the review order, the checklists, the reviewer skill and the prompts work with any agent that can open a browser or read screenshots.

Your front door is any page a stranger judges in a few seconds: a landing page, a pricing page, a portfolio, a product page, a GitHub profile. Coding agents now build these in minutes, and they tend to reach for the same defaults every time. Visitors have learned to spot those defaults, and some stop trusting the page before they read a word. Worse than looking generic, agent-built pages often invent things: testimonials nobody gave, logos of companies that never bought, numbers with no source, a form that says thanks and sends nothing.

This pack is a review gate you run before anything goes live. It has a fixed order (hard floor, truth, usability, machine-made tells, polish), a separate reviewer that sees only screenshots and the checklist, and a browser pass so the agent checks the real page instead of trusting its own code. It is built from public creator breakdowns of AI-made design tells, a UX laws reference, an open-source design taste skill and our own client page reviews. Sources are credited by link in the Sources section.

## What is inside

| Section | What it is for |
|---|---|
| 1. Why a gate, and the order | Five passes, each with its own power to block |
| 2. The hard floor | Nine checks that block publishing, no exceptions |
| 3. The truth pass | Every claim, quote, logo and number traced to evidence |
| 4. UX laws as review questions | 20 well-known usability principles turned into yes or no checks |
| 5. Machine-made tells | 36 common defaults grouped by color, type, layout, parts, copy, trust and motion, each with a fix |
| 6. The browser pass | Real screenshots at three widths, console, forms, links and accessibility |
| 7. The fresh-eyes reviewer | A second session that sees only screenshots and the checklist |
| 8. Decoration discipline | Component libraries and motion without the template look, and why you never clone a site |
| 9. Profile pages | GitHub profiles and about pages as front doors |
| 10. REVIEW.md and the overrule log | One report format and a record of every tell you kept on purpose |
| 11. Rules block | A paragraph for CLAUDE.md, AGENTS.md or a Cursor rule |
| 12. The page-reviewer skill | A SKILL.md that runs the whole gate |
| 13. Prompts | 12 prompts for each pass |
| 14. Sources | What we used and how we used it |
````

## Worth knowing

- A tell is a signal, not a rule. Keep any you have a reason for and log it.
- Automated accessibility scans catch only part of real problems. Do a keyboard walk yourself.
- Not legal advice on testimonials, trade dress or privacy pages.

## Questions

**How is this different from the launch review?** The launch review is a done-for-you check of a whole app. This pack is a gate you run yourself on every page, focused on what visitors see, what is true and what looks generated.

**Do I need to know Playwright?** No. The pack has the prompt that makes your agent write the browser pass script, and the script stays in your repo for every later change.

**Will it make my page look like everyone else's?** The opposite. Most of the tells are the defaults that make pages look alike. The overrule log keeps the ones you chose on purpose.

**Is there a free sample?** Yes. The nine hard floor checks and the browser pass prompt are free on their own listing.

**Is any of it copied?** No. We read 8 public creator breakdowns, references and open-source tools and wrote everything in our own words. Sources are linked and credited inside.

## How to buy

$39 one time at https://paperchasewebb.ai/bots/marketplace/items/front-door-review-stack-page-review-before-publish/. Stripe checkout, then an instant download page; the download link is also in your emailed receipt.

Seller: PaperChaseWebb Inc., mgmt@paperchasewebb.com