prd-draft

Quick Start

What to provide: A feature idea, problem statement, or rough brief. Can be detailed or rough.

/prd-draft                                → Start from scratch with guided questions
/prd-draft [paste your feature idea]      → I'll skip questions you already answered
/prd-draft --stage "team kickoff"         → Set the PRD stage upfront
/prd-draft --ai                           → Include AI behavior specification sections

What you get: A 1-2 page modern PRD with hypothesis, strategic fit, non-goals, success metrics, rollout plan, and optional behavior examples. Saved to thoughts/shared/pm/prds/[feature-name-kebab-case]-[stage].md. Based on thoughts/shared/pm/templates/prd-template.md.

Time: 10-20 minutes for first draft. Then iterate.

Filename convention: [feature-name-kebab-case]-[stage].md (e.g., voice-task-capture-team-kickoff.md)

/prd-draft - Modern PRD Creation

When the PM types /prd-draft, guide them through creating a modern, AI-era PRD.

Context Routing Logic (Internal - for Claude)

Automatic Context Checks: When this skill is invoked, immediately check:

| Source | Files/Folders | Search Terms | What to Extract | | ------------------- | ---------------------------------------------- | --------------------------------- | -------------------------------------------- | | Strategy Docs | thoughts/shared/pm/frameworks/.md | feature name from chat | Strategic pillar alignment | | Related PRDs | thoughts/shared/pm/prds/.md | feature dependencies | Related features and cross-functional impact | | User Research | thoughts/shared/pm/.md | problem related to feature | User pain points, quotes, validation | | Business Model | thoughts/shared/pm/context/business-info-template.md | pricing, revenue, metrics | Revenue impact, North Star alignment | | Competitor Analysis | thoughts/shared/pm/competitive-.md | feature name | Competitive positioning if relevant | | Stakeholder Context | Stakeholder profiles | key stakeholders for this feature | Who to involve, communication style |

Context Priority:

1. Strategic fit and related PRDs FIRST
2. User research and problem statement SECOND
3. Business context and metrics THIRD
4. Competitive positioning FOURTH

Cross-Skill Links:

• If user problem not validated → Link to /interview-guide and /user-research-synthesis
• If impact unclear → Link to /impact-sizing for user/revenue impact
• If success metrics unclear → Link to /feature-metrics for STEDII framework
• If strategic fit unclear → Link to /write-prod-strategy for context

Step 0: Understanding Your Feature Context

Before we draft, let me check what context exists...

Checking:

• thoughts/shared/pm/prds/ for any related feature PRDs
• thoughts/shared/pm/frameworks/ for strategic alignment
• thoughts/shared/pm/ for user validation
• thoughts/shared/pm/context/business-info-template.md for business context
• Stakeholder profiles for who needs to be involved

Context Health Check

After reading context files, verify they contain real data (not placeholder brackets):

If business-info-template.md contains unfilled placeholders (e.g., [Your Company], [Your Product]):

I notice your business context template hasn't been filled out yet. Your PRD will be much stronger with real company context.

Want to:
1. **Fill it out now** (5 min) - I'll ask you the key questions
2. **Proceed without it** - I'll draft the PRD but flag where company context would improve it

If strategy docs are empty/missing: Flag this and suggest /write-prod-strategy or /strategy-sprint first.

Based on what I find, I'll show you:

What We Know About This Feature

Strategic Context:

• [Which strategic pillar does this support? From your strategy doc]
• [Expected user impact: # of users affected]
• [Business outcome: revenue, retention, or engagement impact]

User & Problem Validation:

• [User research confirming this problem: quotes, frequency]
• [Current workaround: how users solve this today]
• [Problem severity: pain level from research]

Feature Dependencies:

• [Related features: what needs to ship first]
• [Cross-functional impact: which teams are involved]
• [Timing constraints: market window, competitive pressure]

PM-Specific Diagnosis Questions

1. Problem Validation: Is this problem validated with real users, or hypothesis?
2. Strategic Clarity: Do you know how this fits into your Q# roadmap?
3. Scope Clarity: Is this a small v1 or full-featured release?
4. Success Metrics: Do you know what success looks like?
5. Execution Readiness: Do engineering and design understand the scope?

When to Use

• Starting a new feature or initiative
• Need to align cross-functional stakeholders
• Moving from prototype to formal spec
• Preparing for planning/prioritization review

How It Works

This is a 3-step conversational workflow:

Step 1: Clarifying Questions

Step 2: Generate First Draft

Step 3: Multi-Agent Review

Step 1: Clarifying Questions

Adaptive Questions Rule: Before asking clarifying questions, check what the PM has already provided in their prompt. Skip questions that are already answered. Start with: "Based on what you shared, I have [X, Y, Z]. Let me confirm: [summary]. A few remaining questions: [only the gaps]."

If the PM provided a detailed brief, you may have enough to skip straight to Step 2. Only ask what's genuinely missing.

When the PM types /prd-draft with no additional context, start with:

Let's create a modern PRD together. I'll reference your company context and PRD template from this workspace.

Before I draft anything, I need to understand the initiative. Don't worry about structure—just talk through it. Use dictation if that's easier.

**Required context:**
1. What problem are we solving? (Be specific about the user pain)
2. What's the hypothesis? (If we build X, then Y will happen because Z)
3. How does this fit our current strategy? (Reference specific goals from business-info.md)
4. What stage is this PRD at? (Team kickoff / Planning review / XFN kickoff / Solution review / Launch readiness / Impact review)

**Important details:**
5. Is this an A/B test or full launch?
6. What are the non-goals? (What are we explicitly NOT doing?)
7. What are the success metrics? (How will we know this worked?)
8. Who are the key stakeholders? (@ mention them if you have profiles in `thoughts/shared/pm/context/stakeholder-template.md` or similar)

**AI-specific (if applicable):**
9. What are example prompts or user inputs?
10. How should it handle edge cases or weird inputs?
11. What should it NEVER do? (Rejection criteria)

Take your time. I'll ask follow-up questions if needed.

Follow-Up Questions

Based on what the PM shares, ask:

• If strategy fit is unclear: "I see this solves [problem], but how does it ladder up to our [specific strategy goal]? Or is this a bet on a new direction?"

• If non-goals are missing: "What are we explicitly NOT including in v1? Any trade-offs you're making?"

• If metrics are vague: "When you say 'increase engagement,' what's the specific metric? What's the target? What's the guardrail (metric we can't harm)?"

• If rollout plan is missing: "Should this be an A/B test first? What's the passing criteria to roll out to everyone?"

• If AI behavior is unclear: "Can you give me 3 example inputs: one that should work great, one that's borderline, and one that should be rejected?"

What NOT to Ask

• Don't ask about technical implementat