Roast My Landing Page

Get a brutally honest audit of your landing page — scored out of 100 with specific fixes.

Onboarding

First-Time Detection

Ask: "Have you used Roast My Landing Page before?"

If new user, show:

Welcome to Roast My Landing Page!

I'll analyze your landing page across 6 dimensions and give you a score
out of 100 — plus specific fixes explained in plain English.

I use established UX and conversion frameworks (Nielsen Heuristics,
StoryBrand, CRO best practices, WCAG accessibility) — not just opinions.

Startup Sequence (Tool Detection + Context Questions)

At the start of every run, do these in parallel:

A. Background Tool Detection (run silently while asking questions):

Search for these tools/MCPs without blocking the user:

Tool	What it does	How to detect
read-website-fast MCP	Reads full page content fast	Check if `mcp__read-website-fast__read_website` tool is available
Playwright CLI skill	Takes screenshots + tests interactions	Check if `playwright-cli` commands are available
Chrome DevTools MCP	Lighthouse audit + console errors	Check if `mcp__chrome-devtools__*` tools are available
Apify MCP	For JS-heavy or blocked pages	Check if `mcp__apify__*` tools are available

B. Context Questions (ask immediately, don't wait for tool detection):

What type of business is this? (SaaS / ecommerce / course / agency / newsletter / other)
Who is your target customer? (1 sentence)
What should visitors DO on this page? (sign up / buy / book demo / download / other)
Is this launched publicly, or still in early access / pre-launch? (This affects how strictly social proof and traction metrics are scored — see references/industry-benchmarks.md for early-stage exceptions.)

Skip any question where the answer is already known from Claude's memory, prior conversations, or information the user provided in their initial message. If ALL answers are already known, confirm them briefly and proceed.

Wait for the user to answer all remaining questions before starting any analysis.

These answers change how the audit is scored — a SaaS page and an ecommerce page have different success criteria. See references/industry-benchmarks.md for details.

Tool Detection Results (after background check completes)

Once background detection finishes AND user has answered context questions:

If all recommended tools are available → Announce tools and proceed to analysis.
If some tools are missing → Tell the user which tools are missing, what they enable, and ask: "Would you like help installing the missing tools? Or should I continue without them?"
- If yes → Provide install commands one by one and guide them through setup:
  - read-website-fast: claude mcp add read-website-fast -s user -- npx -y @just-every/mcp-read-website-fast
  - Playwright CLI: npx skills add lackeyjb/playwright-skill --skill playwright-skill
  - Chrome DevTools: claude mcp add chrome-devtools -- npx -y chrome-devtools-mcp@latest
- If no → Continue, but announce: "Proceeding without [tool]. The report may not cover [what's missing — e.g., 'visual analysis' without Playwright, 'full page content' without read-website-fast]."

Announce to User

Tools available: [list what was found]
Analysis mode: [full visual / text-only]

Blocked URL Handling

Scenario	How to detect	What to say
403/401 (login required)	HTTP status code	"This page requires login. Try: (1) paste the HTML, (2) share a screenshot, (3) use a staging URL without auth."
Cloudflare/bot protection	Challenge page in response	Try Apify RAG Browser. If still blocked: "The page has bot protection. Share a screenshot or paste the HTML."
JavaScript SPA (empty HTML)	Minimal content in response	Use Playwright to render. If unavailable: "I can only see the loading screen. Install Playwright CLI for full analysis."
Timeout (over 15 seconds)	No response	Retry once. Then: "Your page is too slow to reach — and your visitors are having the same problem."
localhost / private IP	URL starts with localhost, 127., 192.168., 10.	"I can't access local URLs. Deploy to a staging URL, or paste the HTML directly."
5xx server error	HTTP 500, 502, 503, 504 status	Retry once after 10 seconds. If still failing: score Performance at 20/100 max. Flag ALL other dimension scores as "provisional — page was returning server errors during analysis." Tell user: "Your page is returning a server error. Fix this first, then re-run for accurate scores."

Agent Teams Roles

When Agent Teams is enabled, run all 6 analyses in parallel:

Role	Agent Name	Dimension
Orchestrator	main	Coordinates all agents, assembles final report, talks to user
First Impression Analyst	audit-impression	3-second test, visual hierarchy, above-fold content
Copy Analyst	audit-copy	Headlines, jargon, benefits vs features, StoryBrand structure
CTA Analyst	audit-cta	Visibility, placement, button copy, contrast, competing CTAs
Trust Analyst	audit-trust	Social proof, credibility markers, risk reversal, objections
Mobile and Accessibility Analyst	audit-mobile	Responsive design, touch targets, WCAG 2.2 AA, font sizes
Technical Analyst	audit-technical	Core Web Vitals, SEO basics, broken links, console errors

When Agent Teams is NOT available: Run all 6 analyses sequentially in the main session. Same depth, just slower.

The 6 Audit Dimensions

Dimension	Weight	What it checks
First Impression	20%	Can a visitor understand what you do in 3 seconds?
Copy and Messaging	20%	Is the writing clear, jargon-free, and benefit-focused?
Call-to-Action	15%	Is the CTA visible, compelling, and the only clear action?
Trust and Social Proof	15%	Are there testimonials, logos, numbers, or guarantees?
Mobile and Accessibility	15%	Does the page work on phones? Is it accessible?
Performance and Technical	15%	Does it load fast? Are SEO basics covered?

Scoring criteria: references/scoring-rubric.md

Frameworks used per dimension:

First Impression + Copy: references/nielsen-heuristics.md, references/storybrand-framework.md, references/copy-analysis-rules.md
CTA + Trust: references/cro-checklist.md
Mobile + Accessibility: references/wcag-essentials.md
Technical: Core Web Vitals (via Lighthouse if Chrome DevTools available)
All dimensions: references/anti-patterns.md (always-wrong patterns to flag immediately)

Industry-specific scoring adjustments: references/industry-benchmarks.md

Analysis Flow

Phase 1: FETCH AND CAPTURE

Fetch page content using best available tool
Take screenshots at 3 viewports if Playwright available (desktop 1440px, tablet 768px, mobile 375px)
Handle blocked URLs per the fallback table above
If page fetched but screenshots unavailable, proceed with text-only analysis

Phase 2: DIMENSION ANALYSIS

Run all 6 dimension analyses (parallel if Agent Teams, sequential otherwise).

Each dimension analysis must:

Score 0-100 based on checkpoints in references/scoring-rubric.md
Record specific findings in this format:

FINDING: [what's wrong — plain English, no jargon]
IMPACT: [why this matters for visitors/business]
FIX:    [specific action to take]
EFFORT: [quick fix (under 30 min) / few hours / redesign needed]

Phase 3: E2E SMOKE TEST (if Playwright available)

Navigate to URL — screenshot above the fold
Scroll full page — screenshot key sections
Click primary CTA — does it work? Where does it go?
Test on mobile viewport (375px) — screenshot
Check console for JavaScript errors
Report: "As a visitor, here's what I experienced..."

Phase 4: SCORING AND REPORT

Collect all 6 dimension scores 1b. If a 5xx error was encountered, mark all scores as provisional and add a disclaimer to the report header: 'NOTE: The page returned a server error during analysis. Scores are provisional and may change when the page is stable. Fix the server error and re-run for accurate results.'
Apply weights to calculate total score (0-100)
Assign letter grade using the scale below
Rank findings by severity: Critical, Warning, Suggestion
Generate "Top 3 Roasts" — the biggest issues, honest but helpful
Generate "Top 3 Wins" — what the page does well
Generate "Path to improvement" — specific steps to go up one letter grade
Output using the format in assets/report-template.md

Phase 4.5: USER REVIEW (before visual report)

After completing the text report (Phase 4), present it to the user and ask:

"Here's the full audit. Would you like to change anything, or should I generate the visual report?"

Wait for the user's response:

If user requests changes → Apply the changes to scores/findings/recommendations, update the text report, and ask again.
If user approves → Proceed to Phase 5 (visual report generation).
If user says nothing specific → Treat as approval and proceed to Phase 5.

Phase 5: VISUAL REPORT (browser preview)

Generate a self-contained HTML report and open it in the user's browser.

Create a JSON object with all audit data (see schema below)
Read the HTML template from assets/report-ui.html
Replace {{REPORT_DATA}} with the JSON string
Save to ./landing-page-roast-[domain]-[date].html
Open in browser: open (macOS), xdg-open (Linux), start (Windows)
Tell the user: "Visual report saved to [path] and opened in your browser."

JSON schema for the report:

{
  "meta": {
    "url": "string",
    "businessType": "string",
    "conversionGoal": "string",
    "targetCustomer": "string",
    "launchStage": "launched|early-access|pre-launch",
    "date": "YYYY-MM-DD",
    "mode": "full visual|text-only",
    "toolsUsed": ["string"],
    "provisional": false
  },
  "overall": {
    "score": 0,
    "grade": "string",
    "verdict": "string"
  },
  "dimensions": [
    { "name": "First Impression", "score": 0, "weight": 0.20, "critical": false },
    { "name": "Copy & Messaging", "score": 0, "weight": 0.20, "critical": false },
    { "name": "Call-to-Action", "score": 0, "weight": 0.15, "critical": false },
    { "name": "Trust & Proof", "score": 0, "weight": 0.15, "critical": false },
    { "name": "Mobile & Access", "score": 0, "weight": 0.15, "critical": false },
    { "name": "Performance", "score": 0, "weight": 0.15, "critical": false }
  ],
  "roasts": [
    { "issue": "string", "why": "string", "fix": "string", "effort": "string" }
  ],
  "wins": [
    { "what": "string", "why": "string" }
  ],
  "findings": {
    "firstImpression": [{ "severity": "CRITICAL|WARNING|SUGGESTION|PASS", "description": "string" }],
    "copyMessaging": [{ "severity": "string", "description": "string" }],
    "callToAction": [{ "severity": "string", "description": "string" }],
    "trustProof": [{ "severity": "string", "description": "string" }],
    "mobileAccess": [{ "severity": "string", "description": "string" }],
    "performance": [{ "severity": "string", "description": "string" }]
  },
  "antiPatterns": [
    { "name": "string", "impact": "string", "severity": "Critical|Warning|Flag" }
  ],
  "improvement": {
    "currentGrade": "string",
    "nextGrade": "string",
    "steps": [
      { "action": "string", "dimension": "string", "currentScore": 0, "projectedScore": 0, "time": "string" }
    ],
    "estimatedTime": "string",
    "projectedScore": 0,
    "projectedGrade": "string"
  },
  "industry": {
    "businessType": "string",
    "benchmark": "string",
    "position": "above|at|below"
  }
}

Grading Scale

Grade	Score	What it means
A+	95-100	Excellent. Minor tweaks only.
A	90-94	Very strong. A few small improvements possible.
B+	85-89	Good page with clear room to improve.
B	80-84	Solid foundation, missing some key elements.
C+	75-79	Average. Several important things need fixing.
C	70-74	Below average. Visitors are probably leaving confused.
D	60-69	Needs serious work. Major issues with clarity or trust.
F	Below 60	Actively hurting your business. Fix urgently.

Output Format

LANDING PAGE ROAST: [url]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Overall Score: [X]/100 ([Grade])
Business: [type] | Goal: [conversion goal]

BREAKDOWN:
  First Impression:  [████████░░]  [score]/100
  Copy and Messaging:[██████░░░░]  [score]/100
  Call-to-Action:    [█████████░]  [score]/100
  Trust and Proof:   [█████░░░░░]  [score]/100
  Mobile and Access: [███████░░░]  [score]/100
  Performance:       [████████░░]  [score]/100

TOP 3 ROASTS:
1. "[issue in plain English]"
   Why it matters: [impact on visitors]
   Fix: [specific action]
   Effort: [time estimate]

2. ...
3. ...

TOP 3 WINS:
1. "[what's working well]"
2. ...
3. ...

YOUR PATH FROM [current grade] TO [next grade]:
1. [specific step] ([dimension]: [current] -> [projected])
2. ...
3. ...
Estimated time: [total hours]
Expected new score: [projected]/100 ([projected grade])

See assets/report-template.md for the full template with all fields.

Report Language Rules

Write every finding in simple English. The user may not be a developer.

DO:

"Your signup button blends into the background — visitors can't find it"
"No testimonials anywhere on the page. Visitors have no reason to trust you yet."
"On phones, the text is too small to read without zooming in"

DON'T:

"CTA has insufficient contrast ratio against the background gradient"
"No social proof elements above the fold impacting conversion rate optimization"
"Font-size below 16px on mobile viewport causes accessibility violations"

Tone: Like a friend who happens to be a UX expert — honest and direct, but always helpful. Never mean-spirited.

Troubleshooting

"Skill can't fetch the page" Check the URL is correct and publicly accessible. Try the fallback chain: WebFetch, then Apify, then ask user to paste HTML.

"No screenshots available" Install Playwright CLI skill for visual analysis. Without it, the audit is text-only (still useful, but misses visual hierarchy and mobile layout issues).

"Score seems too low" Business type affects the benchmark. Make sure the correct business type is set. Security and trust issues are weighted heavily because they directly impact whether visitors convert.

"Score seems too high" The audit checks what's ON the page, not what's missing from the business. A beautiful page with no product-market fit will still score well on UX.

"Analysis is too technical" Report language rules enforce simple English. If a finding slips through with jargon, rephrase it before presenting.

FuzulsFriend/roast-my-landing-page

What is roast-my-landing-page?

Ask in your favorite AI

Documentation