unicorn-page-review-panel: expert page scoring

Review & QA Cowork-ready

What it does

unicorn-page-review-panel is your pre-ship quality gate. Point it at any landing page (live URL or local file) and it assembles 15 core expert disciplines, each channeling a mini-panel of marketing and design legends (Halbert, Schwartz, Ogilvy, Kahneman, Cialdini, Hormozi, Rams, and more). They score the page against its actual objective, weighted by page type, and hand you a ranked fix list.

It always asks for context first: what is this page for, who is it for, where is the traffic from? A score without an objective is just a vibe, and this panel refuses to produce one.

Up to 6 conditional experts load on top of the core 15 when the page needs them: Legal/Compliance for health and finance, SEO authority for content pages, Checkout for ecommerce, and so on.

When to reach for it

You’ve finished a build and want an honest score before it goes live.
You inherited an underperforming page and want a ranked list of what to fix first.
You want to know why a page isn’t converting, with specific evidence instead of generic advice.
You’re running an agency review and need one objective rubric across many pages.
You want to iterate a local file to 90+ without implementing each fix by hand.

Where it runs

Cowork ✓ Desktop ✓

The panel works in both environments. Reviewing a live URL works anywhere. Iterate mode (auto-implement and re-score) needs a local file, so it runs wherever file editing is available.

How to run it

Point it at a page, live URL or local file:

“/unicorn-page-review-panel https://example.com/my-landing-page”

Or describe what you want in plain language:

“Score this advertorial and tell me the top things to fix: output/skincare/advertorial/index.html”

Two modes:

Audit mode (default): read-only review, zero edits, works on live URLs and local files.
Iterate mode (local files only): the panel implements the top fixes, re-scores, and runs up to 3 rounds until the page hits 90+ or the rounds run out. Pass --mode iterate to activate.

What it needs from you

A URL or a local file path. That is the minimum.
(It will ask) The primary objective, if it can’t auto-detect it. “Book a call” vs. “buy the $47 product” changes the entire weighting. Answer this one honestly.
(Optional) Your brand file, if you have one. Without brand context the Brand Strategy expert flags a gap rather than guessing.

What you get back

A structured scorecard and fix list:

Overall score: a single weighted number (90-100 = ship-ready, 80-89 = high-priority fixes needed, below 70 = major issues).
Per-expert scores: each expert returns a score, its top 1-2 findings, and which legends drove the call.
P0/P1/P2 action list: P0 = conversion-blockers to fix before launch, P1 = high-impact improvements, P2 = polish. Every item is attributed to a named legend or framework.
Evidence citations: every finding cites where the issue lives (line number, exact quote, CSS measurement, or screenshot region). No vague recommendations.

A worked example

You: “Score this advertorial and tell me what’s blocking it from shipping: output/creatify/advertorial/index.html”

Panel: (after the context brief)

Overall score: 74 / 100. Objective: get cold-traffic readers to click through to the product page.

Expert Score Top finding
Copywriting 71 Hero headline wraps to 6 lines on mobile. Halbert: a headline’s job is to be read, not admired. Cut to under 10 words.
Trust & Social Proof 68 Three testimonials are display: none on mobile. Cialdini: proof must show at the moment of doubt.
Direct Response 81 Lead and close are strong; the agitate section is thin. Kennedy: buyers must feel the cost of inaction.

P0 (fix before launch):

Hero headline at 375px: 6 lines, cap at 4. Drop to 48px or rewrite. (Halbert, design-standards.md rule #13)

Testimonials hidden on mobile: remove display: none in the 480px breakpoint. (Cialdini)

P1 (high impact):

Agitate section: add one visceral consequence. Not “wasted ad spend” but “another month paying for clicks that don’t convert while your competitor scales.” (Kennedy, Carlton)

Expert	Score	Top finding
Copywriting	71	Hero headline wraps to 6 lines on mobile. Halbert: a headline’s job is to be read, not admired. Cut to under 10 words.
Trust & Social Proof	68	Three testimonials are `display: none` on mobile. Cialdini: proof must show at the moment of doubt.
Direct Response	81	Lead and close are strong; the agitate section is thin. Kennedy: buyers must feel the cost of inaction.

Tips & gotchas

The context brief is not optional. Answer what the page is for. A rubric weighted for “book a call” is completely different from one for “sell a $27 product.” That first answer sets everything.
Audit mode never edits your file. Run it on a live URL with zero risk. It is read-only.
Iterate mode caps at 3 rounds. If the page hasn’t hit 90+ by then, the panel stops and reports what’s left. Some issues need human judgment.
Core Web Vitals scoring needs Chrome MCP (Tier 1 tools). Without it, the Performance expert reports what it can’t measure rather than guessing.
This is not red-team. The panel scores craft: copy, design, CRO, trust, behavioral science. Red-team simulates a hostile buyer talking themselves out of the sale. They pair well: run the panel to reach 90+, then red-team before spending media.
The legends are not decoration. Citing Halbert on a headline or Cialdini on proof placement anchors each fix to a framework you can look up. No attribution means the review isn’t finished.

red-team is the natural companion: panel first (craft quality), red-team second (buyer psychology). apollo feeds the panel when brand or ICP context is thin. page-architect and assessment-builder both call this skill automatically in Phase 3, so anything built with those gets a panel review before it reaches you.