/wizard

---

🚨 EXECUTION NOTICE FOR CLAUDE

When you invoke this command via SlashCommand, the system returns THESE INSTRUCTIONS below.

YOU are the executor. This is NOT an autonomous subprocess.

• ✅ The phases below are YOUR execution checklist
• ✅ YOU must run each phase immediately using tools (Bash, Read, Write, Edit, TodoWrite)
• ✅ Complete ALL phases before considering this command done
• ❌ DON't wait for "the command to complete" - YOU complete it by executing the phases
• ❌ DON't treat this as status output - it IS your instruction set

Immediately after SlashCommand returns, start executing Phase 0, then Phase 1, etc.

See @CLAUDE.md section "SlashCommand Execution - YOU Are The Executor" for detailed explanation.

---

Arguments: $ARGUMENTS

Goal: Conduct interactive wizard to gather requirements, process multimodal inputs, generate architecture docs, and create feature specs ready for implementation.

Core Principles:

• Multimodal first: Accept files, images, URLs, and text
• Progressive disclosure: Ask targeted questions based on inputs
• Comprehensive planning: Generate architecture before specs
• Automation: Script creates structure, agents fill content
• Infrastructure via plugins: No infra specs (plugins handle that)

Phase 1: Welcome and Initial Context Goal: Gather project description and multimodal inputs

Actions:

• Display welcome message
• Create todo list tracking workflow phases
• Ask user for project description and file uploads:
  • "What would you like to build?"
  • "Upload any files (wireframes, docs, code, URLs) to help me understand:"
    • Screenshots/wireframes (PNG, JPG)
    • Requirements docs (PDF, Word, Markdown)
    • Existing code (zip, GitHub URL)
    • Competitor sites (URLs)
• Store initial description in .wizard/initial-request.md
• If user provides file paths or URLs, process them next phase

Phase 2: Process Multimodal Inputs (if provided) Goal: Extract requirements from uploaded materials

Actions:

• If user provided files/images/URLs, invoke requirements-processor:

Task(description="Process uploaded inputs", subagent_type="planning:requirements-processor", prompt="You are the requirements-processor agent. Process all uploaded files, images, and URLs provided by the user.

Inputs to process: $USER_PROVIDED_FILES_AND_URLS

Extract:

• Features and capabilities
• User stories and workflows
• Technical constraints
• Integration requirements
• Data entities
• UI components

Return structured JSON with:

• extracted.features
• extracted.user_stories
• extracted.technical_constraints
• extracted.integrations
• extracted.data_entities
• confidence scores

Deliverable: Complete extraction report in JSON format")

• Save extracted data to .wizard/extracted-requirements.json
• Update todos

Phase 3: Structured Q&A Rounds Goal: Gather additional context through targeted questions

Actions:

• Conduct 6-8 rounds using AskUserQuestion covering:
  • Project type and users
  • Core features (MVP vs. future)
  • Technical stack and integrations
  • Constraints (timeline, budget, team)
  • Success metrics and KPIs
• Save all Q&A to docs/requirements/YYYY-MM-DD-project/02-wizard-qa.md
• Update todos

Phase 4: Generate Architecture (BATCHED PARALLEL) Goal: Create architecture docs using batched parallel agents for speed and UI stability

⚠️ CONSTRAINT: Maximum 10-12 agents per batch (UI breaks with >10)

Actions: Batch 1 (8 agents): Launch architecture + ADRs + roadmap + page specs

• Launch IN PARALLEL (one message, multiple Task calls): Task 1: Generate README.md + backend.md (system overview, Claude Agent SDK, MCP) Task 2: Generate data.md + ai.md (Supabase schema, AI architecture) Task 3: Generate security.md + integrations.md (auth, API keys, external services) Task 4: Generate infrastructure.md + frontend.md (deployment, component patterns) Task 5: Generate application-pages.md (interactive app pages: dashboard, settings, chat, admin) Task 6: Generate website-pages.md (static/marketing pages: landing, pricing, about, blog) Task 7: Generate ADRs (decision-documenter agent) Task 8: Generate ROADMAP.md (roadmap-planner agent)

• Each agent receives same context:

  • Wizard requirements: docs/requirements/
  • Extracted data: .wizard/extracted-requirements.json
  • Q&A: docs/requirements/*/02-wizard-qa.md

• Agents work simultaneously (8 agents, UI-safe batch size)

• Verify all 10 architecture files + ADRs + roadmap created

• Update todos

Phase 4.5: Validate Architecture (CTO Review) Goal: Multi-tier validation of generated architecture

Actions:

• Launch validator agents IN PARALLEL:

  Task 1: Technical validator (checks completeness, diagrams, security) Task 2: Cost validator (verifies budget constraints, estimates costs) Task 3: Timeline validator (confirms 1-3 month aggressive timeline feasible)

• Each validator outputs: validation-report-[type].md

• Launch CTO-level review agent:

  Task: CTO reviewer reads all architecture + validation reports

  • Identifies gaps, inconsistencies, risks
  • Provides executive summary
  • Outputs: docs/architecture/CTO-REVIEW.md

• If critical issues found: Display and ask user to proceed or regenerate

• Update todos

Phase 5: Final Architecture Validation Goal: Validate architecture planning is complete and ready for spec generation

Actions:

• Launch CTO reviewer for architecture approval:

  Task: CTO reviewer reads architecture planning package

  • All 10 architecture files (README.md, backend.md, frontend.md, data.md, ai.md, infrastructure.md, security.md, integrations.md, application-pages.md, website-pages.md)
  • All ADRs and ROADMAP.md
  • project.json (tech stack configuration)
  • features.json (feature breakdown)
  • Previous validation reports (if any warnings)
  • Wizard requirements and Q&A

  CTO validates:

  • Architecture is complete and coherent
  • Tech stack choices are appropriate
  • Features are well-defined and scoped
  • Infrastructure components identified
  • Security considerations documented
  • Plan is ready for spec generation

  Output: docs/FINAL-APPROVAL.md

  • Status: APPROVED | APPROVED_WITH_CHANGES | REJECTED
  • Executive summary of architecture plan
  • Critical issues (blockers)
  • Warnings (should fix before spec generation)
  • Recommendations (optional)
  • Go/no-go decision for proceeding to spec generation

• If REJECTED: Display issues and ask user to regenerate architecture

• If APPROVED_WITH_CHANGES: Display warnings and ask user to proceed or fix

• If APPROVED: Continue to finalization

• Update todos

Phase 6: Finalization Goal: Complete wizard and prepare for spec generation

Actions:

• Update ROADMAP.md with final approval status

• Create summary report:

  • Features defined: X
  • Total estimated time: Y days
  • Architecture docs created:
    • docs/architecture/README.md
    • docs/architecture/backend.md
    • docs/architecture/frontend.md
    • docs/architecture/data.md
    • docs/architecture/ai.md
    • docs/architecture/infrastructure.md
    • docs/architecture/security.md
    • docs/architecture/integrations.md
  • ADRs: Z decisions documented
  • Approval status: APPROVED (from Phase 5)
  • Next steps: Run /planning:extract-config to create project.json and features.json

• Save summary to .wizard/completion-summary.md

• Mark all todos complete

Phase 7: Summary Goal: Display results and next steps

Actions:

• Display completion message:

  ✅ Architecture Planning Complete!
  
  Created Architecture Documentation:
  - docs/requirements/ (wizard Q&A inputs)
  - docs/architecture/README.md (system overview)
  - docs/architecture/backend.md (backend architecture)
  - docs/architecture/frontend.md (frontend architecture)
  - docs/architecture/data.md (database schema)
  - docs/architecture/ai.md (AI stack architecture)
  - docs/architecture/infrastructure.md (infrastructure components)
  - docs/architecture/security.md (security architecture)
  - docs/architecture/integrations.md (external integrations)
  - docs/adr/*.md (architectural decision records)
  - docs/ROADMAP.md (project roadmap)
  
  Next Steps:
  1. Review docs/FINAL-APPROVAL.md for validation results
  2. Run /planning:extract-config to create project.json and features.json
  3. Run /planning:init-project to generate feature specs
  4. Run /foundation:generate-infrastructure-specs for infrastructure specs
  5. Begin implementation following the specs
  

• Update todos to completed