Comprehensive 6-phase SOP for AI-driven landing page creation (Research -> Copy -> Inspiration -> Build -> Iterate -> Deploy). Use when building marketing pages, sales pages, or product landing pages. Combines web research for current best practices, structured copywriting from product briefs, design system extraction via Firecrawl, and multi-model code generation to produce high-converting, deployable landing pages. Integrates with expertise system for marketing/frontend/deployment domains.
Inherits all available tools
Additional assets for this skill
This skill inherits all available tools. When active, it can use any tool Claude has access to.
CHANGELOG.mdexamples/product-brief-template.mdgraphviz/landing-page-process.dotresources/scripts/firecrawl-scraper.jsresources/scripts/netlify-deploy.jsresources/scripts/screenshot-capture.jsname: landing-page-generator version: 2.0.0 description: Comprehensive 6-phase SOP for AI-driven landing page creation (Research -> Copy -> Inspiration -> Build -> Iterate -> Deploy). Use when building marketing pages, sales pages, or product landing pages. Combines web research for current best practices, structured copywriting from product briefs, design system extraction via Firecrawl, and multi-model code generation to produce high-converting, deployable landing pages. Integrates with expertise system for marketing/frontend/deployment domains. category: delivery complexity: medium estimated_time: 2-4 hours triggers:
A comprehensive 6-phase SOP for AI-driven landing page creation, from research to deployment. Converts product/service briefs into high-converting, deployable landing pages using evidence-based design patterns and conversion psychology.
Landing Page Generator represents a systematic approach to creating landing pages that actually convert. Rather than generating generic templates, it employs a research-driven methodology that ensures every page is optimized for its specific audience, product, and conversion goal.
This skill draws inspiration from direct response marketing principles--understanding that a landing page is not just a website, but a carefully crafted argument that guides visitors toward a single action. The 6-phase workflow mirrors how professional conversion rate optimization (CRO) teams work: research the market, craft compelling copy, gather design inspiration, build with purpose, iterate based on feedback, and deploy with confidence.
Key Innovation: This skill combines:
The result is landing pages that don't just look good--they convert visitors into customers.
Landing Page Generator operates on five fundamental principles that ensure every page achieves its conversion goal:
The words on the page matter more than how it looks. A well-written page with mediocre design will outperform a beautiful page with weak copy. This is why Phase 2 (Copy) comes before Phase 3 (Inspiration) and Phase 4 (Build).
In practice:
Every landing page should have exactly one conversion goal. Multiple CTAs, competing messages, or confusing navigation kill conversions. The page exists to drive one specific action.
In practice:
What worked last year may not work today. Landing page best practices evolve constantly. Phase 1 (Research) ensures every page is built on current, evidence-based tactics rather than outdated assumptions.
In practice:
Extracting branding from inspiration sites creates consistency, not copycats. The goal is to capture design systems (colors, fonts, spacing) while maintaining unique value propositions and messaging.
In practice:
AI context windows have limits. After 2-3 iterations, quality degrades as the context fills. Moving to dedicated coding environments (like Cursor) preserves quality during extensive refinement.
In practice:
mcp_servers:
required:
- memory-mcp # Store research, copy, learnings
optional:
- playwright # For screenshot capture
- flow-nexus # For advanced scraping
Why These MCPs:
Before beginning landing page creation:
actions:
- Check: .claude/expertise/marketing.yaml
- Check: .claude/expertise/frontend.yaml
- Check: .claude/expertise/deployment.yaml
if expertise_exists:
- Load: conversion patterns, design systems, deployment configs
- Apply: Proven landing page structures
- Avoid: Known anti-patterns
else:
- Flag: Discovery mode
- Plan: Generate expertise learnings after successful deployment
Use When:
Do Not Use When:
feature-dev-complete)backend-api-development)Different landing page types require different approaches. Identify the type early to optimize the entire workflow:
Patterns: "software", "app", "platform", "tool", "subscription" Common characteristics:
Key research focus:
Copy framework: Problem-Agitation-Solution (PAS)
Patterns: "plumber", "lawyer", "dentist", "contractor", "local" Common characteristics:
Key research focus:
Copy framework: Trust-Authority-Action
Patterns: "buy", "shop", "product", "order", "purchase" Common characteristics:
Key research focus:
Copy framework: Features-Advantages-Benefits (FAB)
Patterns: "contact", "quote", "consultation", "demo", "signup" Common characteristics:
Key research focus:
Copy framework: AIDA (Attention-Interest-Desire-Action)
Patterns: "register", "event", "webinar", "conference", "workshop" Common characteristics:
Key research focus:
Copy framework: Value-Urgency-Action
Patterns: "download", "mobile app", "iOS", "Android", "install" Common characteristics:
Key research focus:
Copy framework: Benefit-Social Proof-Download
Purpose: AI researches current landing page best practices, conversion tactics, and design trends.
Agent: researcher
Prompt Template:
Research the best practices for creating highly converting landing pages as of {TODAY_DATE}.
Focus areas:
- Copywriting tactics (headlines, CTAs, social proof)
- Landing page structure (hero, problem, solution, testimonials, CTA)
- Styling and color psychology
- Mobile responsiveness requirements
- Current conversion rate benchmarks
Include any other areas critical for landing page success.
Output: Render as markdown artifact titled "Landing Page Playbook"
Input Contract:
required:
- today_date: string # Current date for recency
optional:
- industry: string # Industry-specific research
- target_audience: string # Audience demographics
Output Contract:
required:
- playbook_markdown: string # Best practices document
- key_findings: list # Top 5-10 actionable insights
Evidence-Based Techniques:
Purpose: AI writes landing page copy using research insights and product/service brief.
Agent: content-writer or coder
Prompt Template:
Using the Landing Page Playbook from Phase 1, create the copy for my landing page.
IMPORTANT: Only write the TEXT content. Do not write any code yet.
Product/Service Brief:
{PRODUCT_BRIEF}
Include sections:
- Hero headline and subheadline
- Problem statement
- Solution positioning
- Features/Benefits (3-5 key points)
- Social proof (testimonials framework)
- FAQ (3-5 common questions)
- CTA (primary and secondary)
Output: Render as markdown artifact titled "Landing Page Copy"
Input Contract:
required:
- product_brief: string # What you're selling
- playbook: artifact # Phase 1 output
optional:
- tone: string # Professional, casual, urgent
- unique_selling_proposition: string
- pricing: string
- target_pain_points: list
Output Contract:
required:
- copy_markdown: string # All page copy
- section_breakdown: object # Copy organized by section
User Checkpoint: Review copy before proceeding. This is the most critical alignment point.
Purpose: Gather design inspiration from existing websites via branding extraction and screenshots.
Agents: researcher, frontend-performance-optimizer
Two Sub-Tasks:
# Use helper script
node resources/scripts/firecrawl-scraper.js {INSPIRATION_URL}
Script Output: branding.json containing:
# Use helper script
node resources/scripts/screenshot-capture.js {INSPIRATION_URL}
Script Output: inspiration-screenshot.png - Full-page capture for visual reference
Input Contract:
required:
- inspiration_url: string # Website to extract from
optional:
- existing_brand_url: string # Your own site for consistency
- additional_inspiration: list # Multiple URLs
Output Contract:
required:
- branding_json: file # Firecrawl extraction
- screenshot_png: file # Full-page capture
Purpose: AI generates the landing page code using all gathered inputs.
Agent: coder, frontend-dev
Prompt Template:
You are an expert developer. Create an extremely beautiful and well-structured landing page.
I am providing:
1. Landing Page Playbook (best practices) - {PLAYBOOK}
2. Landing Page Copy (all text content) - {COPY}
3. Branding Guidelines (from Firecrawl) - {BRANDING_JSON}
4. Design Inspiration Screenshot - {SCREENSHOT}
Requirements:
- Single HTML file with inline CSS and JavaScript
- Mobile-responsive design
- Use the EXACT copy provided (do not modify text)
- Match branding colors, fonts, and styling from JSON
- Render via artifacts feature to preview
Technical Stack:
- HTML5 semantic markup
- CSS3 with flexbox/grid
- Vanilla JavaScript (no frameworks)
- Optional: Tailwind CSS via CDN
Output: Single index.html file rendered as artifact
Input Contract:
required:
- playbook: artifact # Phase 1 output
- copy: artifact # Phase 2 output
- branding_json: file # Phase 3A output
- screenshot: file # Phase 3B output
optional:
- tech_preferences: object # Framework preferences
- accessibility_level: string # WCAG AA/AAA
Output Contract:
required:
- index_html: file # Complete landing page
- preview_artifact: artifact # Rendered preview
Model Recommendation: Use highest-capability model available:
Purpose: Refine the landing page through targeted iterations using code-focused tooling.
Agent: coder, reviewer
Strategy: After 1-2 iterations in chat, move to dedicated coding environment to prevent context degradation.
In-Chat Iterations (1-2 max):
Make these specific changes to the landing page:
1. {CHANGE_1}
2. {CHANGE_2}
Preserve all existing functionality and copy.
Cursor/External IDE Iterations (3+):
# Download HTML from artifact
# Open in Cursor with browser preview
# Use element selection for targeted edits:
# 1. Open browser preview in Cursor
# 2. Click "Select Element" tool
# 3. Click on section to modify
# 4. Describe changes in chat
Input Contract:
required:
- current_html: file # Current landing page
- changes_requested: list # Specific modifications
optional:
- iteration_number: integer # Track iteration count
- section_focus: string # Specific section to modify
Output Contract:
required:
- updated_html: file # Modified landing page
- changes_log: list # What was changed
Context Preservation: After 2 iterations, export HTML and continue in fresh context or external tool.
Purpose: Deploy the landing page to production via Netlify CLI.
Agent: cicd-engineer, deployment-readiness
Prerequisites:
product.yourdomain.com)Deployment Script:
# Use helper script
node resources/scripts/netlify-deploy.js --site-name {SITE_NAME} --dir {BUILD_DIR}
Manual Netlify CLI Steps:
# Install Netlify CLI
npm install -g netlify-cli
# Authenticate
netlify login
# Initialize site
netlify init
# Deploy
netlify deploy --prod
Input Contract:
required:
- index_html: file # Final landing page
- site_name: string # Netlify site name
optional:
- custom_domain: string # If using custom domain
- environment: string # staging or production
Output Contract:
required:
- deploy_url: string # Live URL
- deploy_id: string # Deployment identifier
- status: string # success or failed
| Phase | Primary Agent | Backup Agent | Capabilities Required |
|---|---|---|---|
| 1. Research | researcher | analyst | web-research, synthesis |
| 2. Copy | content-writer | coder | copywriting, marketing |
| 3. Inspiration | researcher | frontend-dev | scraping, screenshot |
| 4. Build | coder | frontend-dev | HTML/CSS/JS, design |
| 5. Iterate | coder | reviewer | refactoring, review |
| 6. Deploy | cicd-engineer | deployment-readiness | CI/CD, Netlify |
agents:
researcher:
path: claude-code-plugins/ruv-sparc-three-loop-system/agents/foundry/core/researcher.md
type: core-development
subagent_type: researcher
content-writer:
path: claude-code-plugins/ruv-sparc-three-loop-system/agents/foundry/specialized/content-writer.md
type: specialized
subagent_type: content-writer
fallback: coder
coder:
path: claude-code-plugins/ruv-sparc-three-loop-system/agents/foundry/core/coder.md
type: core-development
subagent_type: coder
frontend-dev:
path: claude-code-plugins/ruv-sparc-three-loop-system/agents/foundry/specialized/frontend-dev.md
type: specialized
subagent_type: frontend-dev
reviewer:
path: claude-code-plugins/ruv-sparc-three-loop-system/agents/foundry/core/reviewer.md
type: core-development
subagent_type: reviewer
cicd-engineer:
path: claude-code-plugins/ruv-sparc-three-loop-system/agents/foundry/infrastructure/cicd-engineer.md
type: infrastructure
subagent_type: cicd-engineer
input_contract:
required:
- product_brief: string # What you're selling (name, description, price, value prop)
optional:
- inspiration_url: string # Design inspiration website
- existing_brand_url: string # Your existing website for brand consistency
- tone: string # Professional, casual, urgent, playful
- target_audience: string # Who you're selling to
- industry: string # For industry-specific research
- tech_preferences: object # Framework preferences
- deploy_config: object # Netlify/deployment settings
output_contract:
required:
- landing_page_html: file # Final deployable HTML
- copy_document: artifact # All page copy for reference
- deploy_url: string # Live URL (if deployed)
optional:
- playbook: artifact # Research findings
- branding_json: file # Extracted branding
- iteration_history: list # Change log
- performance_metrics: object # Lighthouse scores
mcp_servers:
required:
- memory-mcp # Store research, copy, learnings
optional:
- playwright # For screenshot capture
- flow-nexus # For advanced scraping
Phase 1 - Store Research Playbook:
memory_store('landing-page/research/lpg-' + Date.now(), playbook, {
// WHO
agent: 'researcher',
agent_category: 'core-development',
capabilities: ['web-research', 'synthesis'],
// WHEN
timestamp: new Date().toISOString(),
unix_timestamp: Date.now(),
// PROJECT
project: 'landing-page-generator',
skill: 'landing-page-generator',
phase: 'phase-1-research',
// WHY
intent: 'research',
purpose: 'Landing page best practices playbook for reuse'
});
Phase 2 - Store Generated Copy:
memory_store('landing-page/copy/lpc-' + Date.now(), copyDocument, {
agent: 'content-writer',
timestamp: new Date().toISOString(),
project: 'landing-page-generator',
intent: 'content',
product_name: productBrief.name,
tone: productBrief.tone || 'professional'
});
Phase 6 - Store Deployment Record:
memory_store('landing-page/deployments/lpd-' + Date.now(), {
url: deployResult.url,
deploy_id: deployResult.deployId,
site_name: siteName
}, {
agent: 'cicd-engineer',
timestamp: new Date().toISOString(),
project: 'landing-page-generator',
intent: 'deployment',
environment: 'production'
});
Retrieve Past Learnings:
// Find similar landing page research
const priorResearch = await vector_search('landing page conversion tactics', {
namespace: 'landing-page/research',
limit: 5
});
// Find past deployments for same product type
const priorDeploys = await vector_search('SaaS landing page', {
namespace: 'landing-page/deployments',
limit: 3
});
Location: resources/scripts/firecrawl-scraper.js
Extracts branding guidelines from any URL using Firecrawl API.
Location: resources/scripts/screenshot-capture.js
Captures full-page screenshots using Puppeteer/Playwright.
Location: resources/scripts/netlify-deploy.js
Automates Netlify CLI deployment with proper configuration.
See: graphviz/landing-page-process.dot
digraph LandingPageGenerator {
rankdir=LR;
node [shape=box, style=filled, fontname="Arial"];
start [shape=ellipse, label="Product Brief", fillcolor=lightgreen];
p1 [label="Phase 1\nResearch", fillcolor=lightblue];
p2 [label="Phase 2\nCopy", fillcolor=lightyellow];
p3 [label="Phase 3\nInspiration", fillcolor=lightcoral];
p4 [label="Phase 4\nBuild", fillcolor=lightpink];
p5 [label="Phase 5\nIterate", fillcolor=lavender];
p6 [label="Phase 6\nDeploy", fillcolor=lightgreen];
end [shape=ellipse, label="Live URL", fillcolor=green, fontcolor=white];
start -> p1;
p1 -> p2 [label="Playbook"];
p2 -> p3 [label="Copy"];
p3 -> p4 [label="Branding\n+Screenshot"];
p4 -> p5 [label="HTML"];
p5 -> p5 [label="Refine", style=dashed];
p5 -> p6 [label="Final HTML"];
p6 -> end;
labelloc="t";
label="Landing Page Generator: 6-Phase SOP (v1.0)";
fontsize=14;
}
// User: "Create a landing page for my AI writing assistant called WriteGenius"
[Single Message]:
Skill("landing-page-generator")
// Phase 1: Research
Task("Researcher", "Research landing page best practices as of 2025-01-15, focus on SaaS AI tools", "researcher")
// Phase 2: Copy (after research)
Task("Content Writer", "Write landing page copy for WriteGenius AI writing assistant, $29/month, helps writers 10x output", "content-writer")
// Phase 3: Inspiration
Task("Frontend Dev", "Extract branding from jasper.ai and capture screenshot", "frontend-dev")
// Phase 4-6: Build, Iterate, Deploy
Task("Coder", "Build landing page using playbook, copy, and branding", "coder")
Task("CI/CD Engineer", "Deploy to Netlify as writegenius-landing", "cicd-engineer")
TodoWrite({ todos: [
{content: "Research SaaS landing page best practices", status: "in_progress", activeForm: "Researching best practices"},
{content: "Write compelling copy for WriteGenius", status: "pending", activeForm: "Writing copy"},
{content: "Extract branding from Jasper.ai", status: "pending", activeForm: "Extracting branding"},
{content: "Capture inspiration screenshot", status: "pending", activeForm: "Capturing screenshot"},
{content: "Build HTML landing page", status: "pending", activeForm: "Building landing page"},
{content: "Review and iterate on design", status: "pending", activeForm: "Iterating on design"},
{content: "Deploy to Netlify", status: "pending", activeForm: "Deploying to Netlify"}
]})
// User: "Build a landing page for my plumbing business, Joe's Plumbing, serving Austin TX"
[Single Message]:
Skill("landing-page-generator")
Task("Researcher", "Research local service business landing page best practices 2025, focus on trust signals and local SEO", "researcher")
Task("Content Writer", "Write copy for Joe's Plumbing Austin TX, emergency service, 24/7, licensed and insured", "content-writer")
Task("Coder", "Build professional local business landing page with contact form, service areas, testimonials", "coder")
TodoWrite({ todos: [
{content: "Research local business landing pages", status: "in_progress", activeForm: "Researching local pages"},
{content: "Write trust-focused copy", status: "pending", activeForm: "Writing copy"},
{content: "Build responsive landing page", status: "pending", activeForm: "Building page"},
{content: "Add contact form integration", status: "pending", activeForm: "Adding contact form"},
{content: "Deploy and test", status: "pending", activeForm: "Deploying"}
]})
Different audiences respond to different approaches:
B2B Audiences:
B2C Audiences:
Technical Audiences:
Non-Technical Audiences:
Different AI models excel at different phases:
| Phase | Recommended Model | Why |
|---|---|---|
| Research | Claude (web search) | Best at synthesis and current info |
| Copy | Claude Opus / GPT-4 | Strong creative writing |
| Build | Claude Opus 4.5 + Extended Thinking | Best design/code generation |
| Iterate | Cursor (any model) | Specialized for code iteration |
Model-Specific Prompting:
Build landing pages with testing in mind:
Version Control:
Testing Elements:
high_impact_tests:
- Headlines (40-50% conversion impact)
- CTA button text and color
- Hero image/video
- Social proof placement
medium_impact_tests:
- Subheadline copy
- Feature order
- Form field count
- Testimonial selection
low_impact_tests:
- Font choices
- Button shape
- Background patterns
- Icon style
Ensure landing pages load fast:
Image Optimization:
Code Optimization:
Target Metrics:
Avoid these common mistakes that kill conversions:
| Anti-Pattern | Problem | Solution |
|---|---|---|
| Feature dumping | Lists features without benefits | Transform every feature into "so you can..." |
| Vague headlines | "Welcome to Our Site" | Specific value prop: "Cut Your Tax Bill in Half" |
| Jargon overload | Industry terms confuse visitors | Write for a smart 12-year-old |
| Weak CTAs | "Submit" or "Click Here" | Action + benefit: "Get My Free Quote" |
| Wall of text | No visual breaks | Short paragraphs, bullets, whitespace |
| Missing urgency | No reason to act now | Scarcity, deadlines, limited availability |
| Anti-Pattern | Problem | Solution |
|---|---|---|
| Navigation overload | Too many exit points | Minimal or no navigation on landing pages |
| Competing CTAs | Multiple equal buttons | One primary CTA, others visually subordinate |
| Stock photo faces | Feels generic, untrustworthy | Real team photos or authentic imagery |
| Tiny mobile text | Unreadable on phones | Minimum 16px body text |
| Auto-playing video | Annoys users, slows load | Click-to-play with compelling thumbnail |
| Below-fold CTA | Users don't scroll | CTA visible above fold AND repeated below |
| Anti-Pattern | Problem | Solution |
|---|---|---|
| Slow load time | Every second costs conversions | Optimize to <3s load time |
| No HTTPS | Trust issues, SEO penalty | Always use SSL |
| Broken mobile | 60%+ traffic is mobile | Mobile-first responsive design |
| No analytics | Can't measure or improve | GA4 + conversion tracking from day 1 |
| Hard-coded content | Can't A/B test | Parameterize key elements |
| Anti-Pattern | Problem | Solution |
|---|---|---|
| Skipping research | Building on assumptions | Always run Phase 1 research |
| Ignoring copy review | AI-generated copy has issues | Human review at Phase 2 checkpoint |
| Generic inspiration | Looks like everyone else | Use YOUR brand or unique inspiration |
| Too many iterations | Context degradation | Export to Cursor after 2-3 rounds |
| No deployment testing | Live site has issues | Always verify deployed URL |
Full 6-Phase Workflow (2-4 hours):
Quick Mode (1-2 hours, skip Phase 1):
Phase 2 Checkpoint (CRITICAL): Before moving to design, verify:
Phase 4 Checkpoint: Before iterating, verify:
Phase 6 Checkpoint: Before considering complete:
| Priority | Approach |
|---|---|
| Speed over quality | Use templates, skip Phase 3 inspiration, minimal iterations |
| Quality over speed | Full research, multiple inspiration sources, 5+ iterations |
| Balanced | Standard 6-phase workflow with 2-3 iterations |
When iterating based on feedback:
Scope feedback clearly:
Batch iterations:
Landing Page Generator works with other skills in the ecosystem:
| Skill | When to Use First | What It Provides |
|---|---|---|
intent-analyzer | Unclear requirements | Clarified landing page goals |
prompt-architect | Complex product briefs | Optimized product description |
research-driven-planning | Large campaign planning | Strategic context for the page |
| Skill | When to Use After | What It Does |
|---|---|---|
testing-quality | Pre-launch validation | E2E tests for forms, links |
performance-optimization | Post-build tuning | Lighthouse score improvement |
seo-specialist | After deployment | On-page SEO optimization |
| Skill | When to Run Together | How They Coordinate |
|---|---|---|
feature-dev-complete | Building app + landing page | Shared branding, linked navigation |
backend-api-development | Forms need backend | API endpoints for form submission |
github-workflow-automation | CI/CD setup | Automated deployment pipeline |
// Full marketing launch workflow
[Sequential]:
// Step 1: Analyze marketing goals
Skill("intent-analyzer")
// Step 2: Build landing page
Skill("landing-page-generator")
// Step 3: Validate before launch
Skill("testing-quality")
// Step 4: Optimize performance
Skill("performance-optimization")
// Step 5: Set up automation
Skill("github-workflow-automation")
Landing Page Generator transforms the complex, multi-disciplinary task of creating high-converting landing pages into a systematic, AI-assisted workflow. By combining research-driven best practices, structured copywriting, design system extraction, and iterative refinement, it produces pages that don't just look professional--they convert visitors into customers.
The 6-phase workflow mirrors how professional CRO teams operate:
Use Landing Page Generator thoughtfully--not every landing page needs the full workflow, but high-stakes campaigns benefit enormously from this systematic approach. The investment in proper research and copy review pays dividends through higher conversion rates and better ROI on marketing spend.
Remember: Copy before design, one page one goal, and always verify at checkpoints. The best landing page is the one that converts.
Landing Page Generator (DELIVERY SKILL)
|
+--> Can be improved BY skill-forge
+--> Generates expertise learnings for marketing domain
+--> Feeds conversion metrics back to research phase
benchmark: landing-page-benchmark-v1
tests:
- lpg-001: SaaS product landing page
- lpg-002: Local business landing page
- lpg-003: Event registration page
minimum_scores:
copy_quality: 0.80
design_fidelity: 0.75
mobile_responsive: 1.00
deploy_success: 1.00
regression: landing-page-regression-v1
tests:
- lpr-001: Copy matches product brief (must_pass)
- lpr-002: Branding extracted correctly (must_pass)
- lpr-003: HTML renders without errors (must_pass)
- lpr-004: Deploys successfully (must_pass)
namespaces:
- landing-page/research/{id}: Research playbooks
- landing-page/copy/{id}: Generated copy
- landing-page/deployments/{id}: Deployment records
- landing-page/metrics: Conversion tracking
- improvement/audits/landing-page-generator: Skill audits
After invoking this skill, verify:
| Metric | Target |
|---|---|
| Copy alignment | 100% matches product brief |
| Design fidelity | Matches inspiration branding |
| Mobile responsive | 100% (test on 3 viewports) |
| Lighthouse Performance | > 90 |
| Lighthouse Accessibility | > 90 |
| Deploy success | Live URL accessible |
| Issue | Solution |
|---|---|
| Firecrawl rate limited | Use backup inspiration method (manual screenshot) |
| Copy doesn't resonate | Return to Phase 2, provide more product context |
| Design looks generic | Provide stronger inspiration URL or existing brand |
| Netlify deploy fails | Check CLI auth, verify site name availability |
| Mobile layout broken | Request specific mobile fixes in Phase 5 |
| Context window exhausted | Export HTML, continue in Cursor IDE or fresh session |
| Artifact rendering fails | Download HTML file directly, open in browser |
| Multiple iterations causing drift | Restart session with fresh context, load only final HTML |
| Model output quality low | Ensure using Opus 4.5 + Extended Thinking or equivalent |
When Phase 5 iterations exceed 2-3 rounds:
# Cursor workflow for extensive iterations
# 1. Open folder: cursor "C:/path/to/landing-page/"
# 2. Open browser preview (globe icon)
# 3. Use element selector for targeted edits
# 4. Each edit = new focused conversation
After successful deployment, extract learnings to improve future landing pages:
post_deployment_learning:
trigger: "Successful deployment with live URL"
extract_learnings:
- conversion_patterns: "What copy/design elements worked"
- technical_patterns: "Build configurations, deployment settings"
- issues_resolved: "Problems encountered and solutions"
propose_expertise_updates:
- domain: marketing.yaml
updates:
- "Effective headline patterns for {industry}"
- "CTA placement that converted"
validation:
- Run adversarial validation (expertise-adversary)
- Survival threshold: 70%
- Apply only if validated
storage:
namespace: "landing-page/learnings/{deployment_id}"
retention: "permanent"
Remember: The copy (Phase 2) is the most critical checkpoint. Get user approval before investing in design and development phases.