Prd

/omg-prd - Complete Product Documentation Generator

Command: /omg-prd Skill: prd.md Mode: Plan Mode → Execution

Command Behavior

When user types /prd, invoke the prd skill:

Use the prd skill to generate complete 11-file documentation suite for AI agent implementation.

This will:

1. Check for existing VISION.md (from /vision)
2. Ask clarifying questions if needed
3. Generate all 11 documentation files
4. Save to /docs folder
5. Provide implementation roadmap

Arguments

# Standard (after /vision - recommended)
/prd

# From scratch (asks all questions)
/prd --new

# From existing brief/doc
/prd --input /path/to/brief.md

# Quick mode (MVP only, minimal docs)
/prd --quick

# Platform-specific
/prd --platform web
/prd --platform mobile
/prd --platform desktop

# Skip certain files
/prd --skip-testing
/prd --skip-design-system

# Verbose mode (more detailed specs)
/prd --verbose

Output Structure

Generates /docs folder with:

/docs
├── VISION.md                    # Product identity (from /vision or generated)
├── PRD.md                       # Complete product requirements
├── TECH-ARCHITECTURE.md         # Technical architecture
├── DESIGN-SYSTEM.md             # Design system & visual language
├── LAYOUTS.md                   # Screen layouts & navigation
├── DATA-MODEL.md                # Complete data architecture
├── AGENT-PLAYBOOK.md            # AI agent implementation guide
├── TESTING-STRATEGY.md          # Testing plan
├── CHANGELOG-TEMPLATE.md        # Living documentation
├── GLOSSARY.md                  # Project terminology
└── FEATURES/                    # Individual feature specs
    ├── F-001-feature-name.md
    ├── F-002-feature-name.md
    └── ...

Workflow Integration

With /build (Recommended - Full Pipeline)

# Full pipeline: /vision -> /prd -> /brand-identity -> create a plan -> execution
/build

With /vision (Manual Step-by-Step)

# Step 1: Define emotional foundation
/vision

# Step 2: Generate complete docs from vision
/prd

# Step 3: Brand identity system
/brand-identity

# Step 4: Plan with context-aware steps
create a plan

# Step 5: Implement with agents
/team

Standalone

# Generate from scratch
/prd --new

With existing brief

# Convert existing doc to full suite
/prd --input PROJECT_BRIEF.md

What Gets Generated

Core Documents (always)

• VISION.md
• PRD.md
• TECH-ARCHITECTURE.md
• AGENT-PLAYBOOK.md
• DATA-MODEL.md

Design Documents (unless --skip-design-system)

• DESIGN-SYSTEM.md
• LAYOUTS.md

Testing Documents (unless --skip-testing)

• TESTING-STRATEGY.md

Meta Documents (always)

• CHANGELOG-TEMPLATE.md
• GLOSSARY.md

Feature Specs (based on feature count)

• FEATURES/F-XXX-name.md for each P0 and P1 feature

Discovery Questions

If VISION.md doesn't exist, will ask:

Foundation:

1. Project name and one-liner?
2. Core problem (emotional)?
3. Target user mindset?
4. Desired experience?
5. Anti-positioning?

Technical: 6. Platform (web/mobile/desktop)? 7. Tech stack preference? 8. Business model? 9. Auth needed? 10. Payments needed?

Scope: 11. MVP features (P0)? 12. Nice-to-have (P1)? 13. Post-MVP (P2-P3)?

Design: 14. Design references? 15. Aesthetic preference? 16. Color/branding?

Constraints: 17. Budget? 18. Timeline? 19. Compliance? 20. Technical limitations?

Integrations: 21. Third-party services? 22. Existing systems? 23. Analytics?

Quality Gates

Before finalizing, verifies:

Completeness:

• All 11 files generated
• All P0 features have specs
• GUARDIAN: Feature count in PRD.md (P0+P1) == file count in FEATURES/ ← NON-NEGOTIABLE
• All tech choices justified
• All dependencies mapped

Specificity:

• No vague requirements
• All acceptance criteria testable
• All API endpoints specified
• All data fields typed

Agent-Ready:

• AGENT-PLAYBOOK.md has ordered tasks
• Each task has exact file paths
• Gotchas documented
• Validation steps provided

Cross-References:

• Files reference each other correctly
• Feature IDs consistent across files
• Screen IDs match layouts
• Task IDs match dependencies

Example Session

User: /prd

Claude checks for VISION.md:

Found VISION.md from /vision!
Using "TapeLog - Voice Journal" as foundation.

I'll generate the complete 11-file documentation suite.
Starting with technical questions...

Platform: You mentioned mobile. iOS only, or cross-platform?

User: "iOS only for MVP"

Claude:

Good. Tech stack recommendation:
- React Native + Expo (fast iteration, good audio support)
- Supabase (backend, auth, storage)
- Clerk (if you want premium auth UX)

Sound good, or do you have preferences?

User: "React Native + Supabase sounds great"

Claude:

Perfect. Now let's define MVP scope.

From VISION.md, core experience is "pressing record on a cassette deck".

MVP P0 features:
1. Record audio entry
2. Playback entries
3. Organize by date
4. Simple auth (email)

Any others critical for launch?

User: "That's it"

Claude:

Generating documentation suite...

✅ VISION.md (already exists)
✅ PRD.md (4 P0 features, 3 user journeys, success metrics)
✅ TECH-ARCHITECTURE.md (React Native + Expo + Supabase)
✅ DATA-MODEL.md (Entry, User tables)
✅ FEATURES/ (4 files)
✅ DESIGN-SYSTEM.md (vintage cassette aesthetic)
✅ LAYOUTS.md (3 screens: Record, Library, Entry Detail)
✅ AGENT-PLAYBOOK.md (18 ordered tasks)
✅ TESTING-STRATEGY.md
✅ CHANGELOG-TEMPLATE.md
✅ GLOSSARY.md

Documentation complete! 📚

Next steps:
1. Review /docs folder
2. Run /team team build to implement
3. Or manually assign tasks to agents

Integration with /team

After generating docs:

# Option 1: Team Build (recommended)
/team team build

# Agents will:
# - Read AGENT-PLAYBOOK.md for tasks
# - Reference TECH-ARCHITECTURE.md for setup
# - Follow DESIGN-SYSTEM.md for UI
# - Use FEATURES/ for detailed specs
# - Implement in parallel with no ambiguity

# Option 2: Manual
# Review /docs and assign tasks yourself

Advanced Usage

Custom Tech Stack

/prd --stack "Next.js,Convex,Clerk,Stripe"

Specific Compliance

/prd --compliance hipaa
/prd --compliance gdpr
/prd --compliance sox

Multi-Language

/prd --i18n en,fr,es

Performance-Critical

/prd --perf-target "load<1s,interaction<100ms"

Troubleshooting

Q: VISION.md not found, but I ran /vision? A: Check if VISION.md is in project root. If not, /prd will generate it.

Q: Too many features, docs are huge? A: Use --skip-research for MVP only, or manually trim P1/P2 features after generation.

Q: Want to regenerate single file? A: Edit the file manually, or delete it and run /prd --update to regenerate missing files.

Q: Agents not following specs? A: Check AGENT-PLAYBOOK.md has concrete file paths and gotchas. Update and re-run agents.

Output Location

Default: /docs folder in project root

Custom:

/prd --output /path/to/custom/location

Git Integration:

• Automatically creates .gitignore entry for temp files
• Commits /docs folder
• Adds PR template referencing VISION.md

See Also

• /vision - Create VISION.md first (was /branding)
• /build - Full pipeline: /vision + /prd + /brand-identity + create a plan + execution
• /team - Complete product workflow
• /team - Implement from docs with parallel agent teams
• Skill file: $HOME/.claude/skills/prd.md
• Methodology: ~/projects/work/team/docs/prdide.md

Last Updated: 2026-02-13 Output: 11-file documentation suite in /docs