feat(skills): add aidd-landing-page skill

AGENTS.md

@@ -51,3 +51,4 @@ and read `aidd-custom/config.yml` to load configuration into context.
 ## Task Index
 
 fix bug => /aidd-fix
+landing page => /aidd-landing-page

ai/skills/aidd-landing-page/SKILL.md

@@ -0,0 +1,143 @@
+---
+name: aidd-landing-page
+description: >
+  Build high-converting landing pages with clear value propositions, focused CTAs,
+  and measurable performance. Use when planning, building, or reviewing landing pages,
+  conversion funnels, or marketing pages.
+compatibility: Requires a web framework (Next.js, etc.), a deployment target, and optionally a Google PageSpeed Insights API key for automated performance audits.
+---
+
+# Landing Page Engineer
+
+Act as a top-tier conversion-focused front-end engineer and UX strategist.
+Your job is to build landing pages that convert visitors into users by
+combining clear messaging, focused design, and measurable performance.
+
+Capabilities {
+  conversion rate optimization (CRO)
+  value proposition design
+  copywriting for clarity and action
+  visual hierarchy and layout
+  performance optimization (Core Web Vitals)
+  A/B test planning
+  social proof strategy
+  objection handling
+  accessibility
+}
+
+Constraints {
+  One page = one goal. Never split visitor attention across competing CTAs.
+  The headline must explain what the product does — not how it feels. A visitor should understand the value proposition within 3 seconds.
+  Above-the-fold is for the value proposition + primary CTA. Hero images, abstract shapes, and decorative whitespace do not sell — remove them unless they reinforce the message.
+  Show the product early. Real screenshots, dashboards, or interface previews build trust. Concepts and illustrations do not.
+  CTAs must be specific. "Get Started" is weak. "Start free 14-day trial" tells users exactly what happens next.
+  Social proof must reduce perceived risk. Logos alone are insufficient — testimonials citing real, measurable results convert better.
+  Remove anything that makes users think. Excess sections, gratuitous animations, and unclear icons slow decision-making. Clarity beats creativity.
+  Answer objections before visitors ask. Pricing, integrations, security, and support must be visible — hiding them lowers trust.
+  Speed is part of design. Target Lighthouse Performance score ≥ 90 and Largest Contentful Paint < 2.5 s. A beautiful page that loads in 4 seconds is a bad page.
+  Design hierarchy must guide the eye. If everything is bold, colorful, and animated, nothing is important. Use size, contrast, and whitespace to create a single visual path from headline → proof → CTA.
+  Use /aidd-ui for component design, styling, and accessibility.
+  Use /aidd-stack for framework and deployment decisions.
+  Communicate each step to the user as friendly markdown prose — not raw SudoLang syntax.
+}
+
+## Step 1 — Discover
+
+discover(brief) => pageSpec {
+  1. Clarify the single conversion goal (signup, waitlist, purchase, demo, etc.)
+  2. Identify target audience, key objections, and competitive differentiators
+  3. Draft the value proposition: headline, subhead, and primary CTA label
+  4. List social proof assets available (testimonials, logos, metrics, case studies)
+  5. Produce a pageSpec: goal, audience, value prop, proof inventory, objection map
+
+  Constraints {
+    No implementation yet — discovery only
+    Value proposition must pass the 3-second clarity test
+    CTA label must describe the concrete next step
+  }
+}
+
+## Step 2 — Plan Sections
+
+planSections(pageSpec) => sectionPlan {
+  1. Map the visual flow: headline → social proof → product preview → benefits → objection handling → final CTA
+  2. For each section, specify: purpose, copy direction, required assets
+  3. Validate that every section serves the single conversion goal — remove any that don't
+  4. Define responsive breakpoints and above-the-fold content for mobile + desktop
+  5. Produce a sectionPlan document
+
+  Constraints {
+    Fewer sections > more sections
+    Every section must earn its place by advancing the visitor toward the CTA
+    No section should introduce a competing goal
+  }
+}
+
+## Step 3 — Build
+
+build(sectionPlan) => landingPage {
+  Using /execute:
+  1. Scaffold the page structure using the project's framework and component library
+  2. Implement sections top-down following the sectionPlan
+  3. Optimize images: use next-gen formats (WebP/AVIF), lazy-load below-the-fold assets, and size hero images to the rendered dimensions
+  4. Ensure semantic HTML, proper heading hierarchy, and ARIA labels for accessibility
+  5. Run /audit-speed and iterate until targets are met
+
+  Constraints {
+    Inline critical CSS or use framework-native optimizations to eliminate render-blocking resources
+    No JavaScript-driven layout shifts — all dimensions must be explicit or reserved
+    Animations must be CSS-only, respect prefers-reduced-motion, and never delay content visibility
+  }
+}
+
+## Step 4 — Audit Speed
+
+auditSpeed(url | localServer) => performanceReport {
+  Prefer the PageSpeed Insights REST API (no API key required for light usage):
+
+  ```
+  GET https://www.googleapis.com/pagespeedonline/v5/runPagespeed
+    ?url={url}&category=performance&category=accessibility&category=seo
+  ```
+
+  1. Fetch the report; extract Performance score, LCP, CLS, TBT
+  2. Flag any metric that misses the target:
+     - Performance score < 90
+     - LCP > 2.5 s
+     - CLS > 0.1
+     - TBT > 200 ms
+  3. List specific audit failures and recommended fixes
+  4. Re-run after fixes to confirm improvement
+
+  localServer => use Lighthouse CLI (`npx lighthouse {url} --output json --quiet`) if the page is not publicly accessible
+
+  Constraints {
+    Never ship a page with Performance score < 90 without explicit user approval
+    Treat accessibility score < 90 as a blocking issue
+  }
+}
+
+## Step 5 — Review
+
+review(landingPage) => reviewedPage {
+  1. Run /review on all changed files
+  2. Walk through the page as a first-time visitor — does the value prop land in 3 seconds?
+  3. Verify the single-goal constraint: count distinct CTAs. More than one primary action => fix.
+  4. Check objection coverage: pricing, integrations, security, support all visible?
+  5. Confirm social proof includes measurable results, not just logos
+  6. Run /audit-speed one final time
+  7. Resolve all issues found
+
+  Constraints {
+    Every review finding must be resolved before shipping
+  }
+}
+
+landingPage = discover |> planSections |> build |> auditSpeed |> review
+
+Commands {
+  🚀 /landing-plan  - discover + plan sections for a landing page
+  🏗️ /landing-build - build the landing page from an approved section plan
+  🔍 /landing-review - review an existing landing page against all constraints
+  ⚡ /audit-speed   - run a PageSpeed Insights audit and report results
+}

plan.md

@@ -22,6 +22,12 @@
 **Goal**: Install Context7 MCP server to provide agents access to up-to-date documentation for popular libraries, frameworks, and APIs  
 **Tasks**: 2 tasks (API key gathering, installation execution)
 
+### 📋 Landing Page Skill Epic
+
+**Status**: 📋 PLANNED  
+**File**: [`tasks/landing-page-skill-epic.md`](./tasks/landing-page-skill-epic.md)  
+**Goal**: Add an `aidd-landing-page` skill that guides agents to build high-converting landing pages with clear messaging, focused CTAs, and measurable performance
+
 ### 📋 Context7 GitHub Action Epic
 
 **Status**: 📋 PLANNED  

tasks/landing-page-skill-epic.md

@@ -0,0 +1,84 @@
+# Landing Page Skill Epic
+
+**Status**: 📋 PLANNED
+**Goal**: Add an `aidd-landing-page` skill that guides agents to build high-converting landing pages with clear messaging, focused CTAs, and measurable performance
+
+## Overview
+
+Agents currently lack structured guidance for building landing pages. Without conversion-focused constraints, generated pages tend toward decorative layouts, vague headlines, and competing calls-to-action — all of which hurt conversion. This skill codifies proven landing page principles as machine-readable constraints so agents produce pages that convert by default.
+
+---
+
+## Skill File
+
+Create the `aidd-landing-page` skill following the AgentSkills.io spec and the project's constraint-based programming style.
+
+**Requirements**:
+- Given an agent is asked to build a landing page, should have a discoverable skill file at `.cursor/skills/aidd-landing-page/SKILL.md` with valid AgentSkills.io frontmatter
+- Given the skill is loaded, should define a role preamble, Capabilities block, and Constraints block that encode all ten landing page principles
+- Given the skill defines commands, should expose `/landing-plan`, `/landing-build`, `/landing-review`, and `/audit-speed` subcommands
+
+---
+
+## Conversion Constraints
+
+Encode the core landing page rules as enforceable constraints.
+
+**Requirements**:
+- Given a landing page headline, should explain what the product does so a visitor understands the value proposition within 3 seconds
+- Given a landing page, should have exactly one conversion goal — no competing CTAs
+- Given above-the-fold content, should contain the value proposition and primary CTA — not decorative imagery
+- Given a landing page, should show real product screenshots or interface previews early in the page
+- Given a CTA label, should describe the concrete next step rather than a generic phrase like "Get Started"
+- Given social proof content, should include testimonials with measurable results — not logos alone
+- Given page sections, should each directly advance the visitor toward the single conversion goal — remove any that don't
+- Given a landing page, should visibly address common objections (pricing, integrations, security, support) without requiring the visitor to search
+- Given design hierarchy, should use size, contrast, and whitespace to create a single visual path from headline to proof to CTA
+
+---
+
+## Performance Audit
+
+Integrate a PageSpeed Insights audit step into the skill workflow.
+
+**Requirements**:
+- Given a deployed or locally-served landing page URL, should fetch a PageSpeed Insights report and extract Performance score, LCP, CLS, and TBT
+- Given a Performance score below 90 or LCP above 2.5 s, should flag the metric and list specific remediation steps
+- Given a page that is not publicly accessible, should fall back to Lighthouse CLI for local auditing
+- Given an accessibility score below 90, should treat it as a blocking issue
+
+---
+
+## Discover and Plan Workflow
+
+Define a structured discovery-to-plan flow before any implementation begins.
+
+**Requirements**:
+- Given a landing page brief, should produce a pageSpec covering: single conversion goal, target audience, value proposition, social proof inventory, and objection map
+- Given an approved pageSpec, should produce a sectionPlan that maps the visual flow from headline through final CTA
+- Given a sectionPlan, should validate that every section serves the single conversion goal and remove any that don't
+
+---
+
+## Build Workflow
+
+Guide the agent through implementation with performance and accessibility guardrails.
+
+**Requirements**:
+- Given a sectionPlan, should scaffold and implement the page top-down using the project's framework and component library
+- Given image assets, should optimize to next-gen formats, lazy-load below-the-fold images, and size hero images to rendered dimensions
+- Given animations, should be CSS-only, respect `prefers-reduced-motion`, and never delay content visibility
+- Given the build is complete, should automatically run the speed audit and iterate until performance targets are met
+
+---
+
+## Review Workflow
+
+Provide a structured review checklist that validates all constraints.
+
+**Requirements**:
+- Given a completed landing page, should walk through as a first-time visitor and verify the value proposition lands in 3 seconds
+- Given the review step, should count distinct primary CTAs and flag any page with more than one
+- Given the review step, should verify objection coverage for pricing, integrations, security, and support
+- Given the review step, should confirm social proof includes measurable results
+- Given the review step, should run a final speed audit and resolve all findings before shipping