/init-website | planning | ClaudePluginHub

Arguments: $ARGUMENTS

Goal: Analyze comprehensive website description and generate ALL website page specifications (W001-W0XX) in parallel. Creates complete planning documentation for marketing/content website.

Phase 1: Parse Input Goal: Determine input mode and basic context

Actions:

Create todo: "Initialize website specs from description"
Parse $ARGUMENTS:
- If contains "--doc=": MODE = "document", extract DOC_PATH
- Otherwise: MODE = "text", DESCRIPTION = $ARGUMENTS
If MODE = "document":
- Validate file exists: !{bash test -f "$DOC_PATH" && echo "exists" || echo "missing"}
- If missing: Error and exit
Display: "Mode: [MODE]"

Phase 2: Load Project Context Goal: Understand existing project structure

Actions:

Read configuration files:
- @.claude/project.json (tech stack - should have Astro/website framework)
- @website-design.json (existing website pages if any)
- @features.json (to avoid overlap - features are separate from website)
Check if website directory structure exists:
- !{bash test -d specs/website && echo "exists" || echo "create"}
Display: "Project context loaded"

Phase 3: Analyze Website Description Goal: Break down website into discrete pages

Actions:

Launch the feature-analyzer agent to analyze website description:

Task( description="Analyze website and break into pages", subagent_type="planning:feature-analyzer", prompt="Analyze this website description and break it into discrete PAGES (not features).

Input Mode: [MODE] Description: $ARGUMENTS Document Path: [DOC_PATH if applicable]

Read schema template:

@~/.claude/plugins/marketplaces/dev-lifecycle-marketplace/schemas/website-design-schema.json

Analyze:

Read .claude/project.json for tech stack (should have Astro or website framework)
Read website-design.json for existing pages
Identify distinct website pages needed:
- Landing page, pricing, about, blog, docs, contact, etc.
- NOT application features (those go in features.json)
- Marketing/content pages only
For each page:
- Assign W00X ID (sequential: W001, W002, W003...)
- Determine route (/, /pricing, /about, etc.)
- Identify page type (landing, content, interactive, blog)
- Extract content requirements
- Determine components needed
Check for duplicates with existing website-design.json pages

Return JSON array of pages: [ { 'page_id': 'W001', 'route': '/', 'title': 'Landing Page', 'type': 'landing', 'description': 'Main landing page with hero, features, CTA', 'components': ['Hero', 'Features', 'Testimonials', 'CTA'], 'content_needs': ['Hero copy', 'Feature list', 'Social proof'] }, { 'page_id': 'W002', 'route': '/pricing', 'title': 'Pricing', 'type': 'content', 'description': 'Pricing tiers and comparison', 'components': ['PricingTable', 'FAQ', 'CTA'], 'content_needs': ['Pricing tiers', 'FAQ content', 'Feature comparison'] } ]" )

Parse agent response (JSON array of pages)
Display: "Identified X website pages"

Phase 4: Update website-design.json Goal: Register all pages before generating specs

Actions:

Read website-design.json (or create if missing)
For each page from Phase 3:
- Add entry with page_id, route, title, type, description, components
- Record created timestamp
Write updated website-design.json
Display: "✅ Registered X pages in website-design.json"

Phase 5: Generate Page Specs in Parallel Goal: Create all page specifications simultaneously

Actions:

For each page from Phase 3, spawn feature-spec-writer agent
Send ALL Task calls in ONE message (parallel execution)
Provide each agent with: page_id, route, title, type, description, components, content_needs
Each agent creates specs/website/[PAGE_ID]-[slug]/spec.md and tasks.md
Wait for all agents to complete
Display: "✅ Generated specs for X pages"

Phase 6: Summary Goal: Report results and next steps

Actions:

Mark todo complete
Display: "✅ Website initialized:"
- Pages: X pages defined
- Specs: specs/website/W001-..., W002-..., etc.
- Config: website-design.json updated
Next steps:
- Review specs: specs/website/
- Implement pages: /website-builder:* commands
- Generate content: /website-builder:generate-content
- Build site: /website-builder:deploy-marketing-site

Important Notes:

Website vs Features:

Website pages (W00X) = Marketing/content site (Astro, static)
Features (F00X) = Application functionality (Next.js, dynamic)
Keep these SEPARATE - don't mix them

Spec Location: All website page specs go in specs/website/W00X-slug/

Implementation: Use /website-builder:* or /nextjs-frontend:* commands to build pages from specs