| name | prd |
| description | Use this skill when the user asks to "write a PRD", "create a product spec", "document product requirements", "update my PRD", "write a product requirements document", "spec out a feature", or needs to create or update the central document that communicates what to build and why to design and engineering partners.
|
| version | 1.0.0 |
PRD (Product Requirements Document)
Generate or update a PRD — the single source of truth that communicates what to build, why, for whom, and how success will be measured.
When to Use
- After opportunity validation: User value and business value validated, ready to specify
- After feature design: Update with final design decisions, prototypes, risks
- From scratch: Guide the PM through the required thinking even without prior artifacts
Process
Step 1: Gather Inputs
Check workspace for existing artifacts:
- Manager briefing (strategic fit, hypotheses)
- User Value Map or interview synthesis
- Funnel analysis or business value analysis
- Design prototypes or mockups
- Existing draft PRD to update
If artifacts exist, pull relevant sections. If nothing exists, guide from scratch.
Also ask: Output format: What format for the final document? (docx, md, or pdf). Default: docx.
Step 2: Build Section by Section
Walk through interactively. Challenge weak or vague answers.
Section 1: Problem Statement
- What is the problem? (user's perspective)
- Who has it? (specific audience)
- What happens today without this feature? (pain, workarounds)
- What evidence supports this? (data, interviews, tickets)
Push back if the "problem" is a solution in disguise.
Section 2: Target Audience
- Primary user: specific demographics/firmographics + product behaviors
- Secondary users: anyone else affected
- Anti-audience: who is NOT the target for v0
Section 3: Strategic Context
- Connection to company/product/team goals
- Why now? Trigger or urgency?
- What happens if we don't build this?
Section 4: Proposed Solution
- High-level WHAT (not HOW it's built)
- Key user flows step by step
- Entry points: where/how users access
- Interaction with existing features
- Prototypes or mockups (if available)
Section 5: Requirements
- Functional: What the feature must do (user-facing behaviors)
- Non-functional: Performance, security, accessibility, scalability
- Edge cases: Unusual scenarios
Each requirement gets: priority (must-have vs. nice-to-have) and acceptance criteria.
Section 6: Success Metrics
- Primary metric: THE one metric for success
- Supporting metrics: Additional signals
- Guardrail metrics: What should NOT get worse
- How and when to measure
Section 7: Scope & Anti-scope
- In scope (v0): Closed, specific list
- Anti-scope: What does NOT ship (be aggressive)
- Future considerations: Intentionally deferred
Section 8: Design Decisions & Trade-offs
- Alternatives considered and why rejected
- Key trade-offs made
- What was prototyped, tested, learned
Section 9: Risks
- Usability risks (friction, adoption)
- Utility risks (might not solve the problem)
- Technical risks (complexity, dependencies)
- Business risks (timing, competition)
- Each with mitigation plan
Section 10: Open Questions
- Unresolved items needing input
- Blocked decisions and by whom
Step 3: Consistency Check
Before generating the final document, validate the PRD against prior artifacts. This step catches contradictions before they reach design and engineering.
If a manager briefing exists, check:
- Does the PRD's problem statement match the user value hypothesis from the briefing?
- Are the success metrics aligned with the business value the manager identified?
- Does the strategic context in the PRD match the strategic fit from the briefing?
- If there are mismatches, they're not necessarily wrong — the PM may have learned something new. But flag them explicitly: "Your PRD says [X], but the manager briefing said [Y]. Is this an intentional update based on what you learned, or an inconsistency to resolve?"
If interview synthesis or User Value Map exists, check:
- Does the problem statement reflect what users actually said, or what the PM hoped they'd say?
- Is the target audience consistent with where the strongest signal came from in the interviews?
- Are there interview findings that contradict the proposed solution? (e.g., users described the problem differently than the PRD frames it)
- Did the interviews surface severity evidence that should appear in the PRD's strategic context?
If experiment results exist, check:
- Do the success metrics account for learnings from past experiments?
- Is the proposed solution consistent with what was validated?
Internal consistency check (always run, even without prior artifacts):
- Does every functional requirement trace back to the problem statement? If a requirement doesn't connect to the stated problem, it's either scope creep or the problem statement is incomplete.
- Do the success metrics actually measure whether the problem was solved? (Not just whether the feature was shipped or used)
- Is anything in "anti-scope" that contradicts a must-have requirement?
- Are there risks listed without mitigation plans?
Present all inconsistencies to the PM before generating the document. Let them resolve each one — update the PRD, update the prior artifact, or acknowledge the deliberate change.
Step 4: Generate the Document
Produce the document in the user's chosen format (default: docx). If docx, use the docx skill. If pdf, use the pdf skill. If md, write as markdown. Use the following structure:
# PRD: [Feature Name]
Author: [PM name]
Date: [date]
Status: [Draft / In Review / Approved]
## 1. Problem Statement
[problem + audience + evidence]
## 2. Target Audience
### Primary
[who]
### Secondary
[who]
### Anti-audience
[who is NOT the target]
## 3. Strategic Context
[goals + why now]
## 4. Proposed Solution
### Overview
[high-level description]
### Key User Flows
[step-by-step]
### Entry Points
[access points]
### Interactions
[what it touches]
## 5. Requirements
### Functional
| # | Requirement | Priority | Acceptance Criteria |
|---|------------|----------|-------------------|
| 1 | [req] | Must-have | [criteria] |
### Non-functional
[performance, security, accessibility]
### Edge Cases
[unusual scenarios]
## 6. Success Metrics
| Metric | Type | Target | Measurement |
|--------|------|--------|------------|
| [metric] | Primary | [target] | [how] |
| [metric] | Guardrail | [threshold] | [how] |
## 7. Scope
### In Scope (v0)
- [item]
### Anti-scope
- [item]
### Future
- [item]
## 8. Design Decisions & Trade-offs
[alternatives, trade-offs, test results]
## 9. Risks
| Risk | Type | Severity | Mitigation |
|------|------|----------|-----------|
| [risk] | [type] | [H/M/L] | [plan] |
## 10. Open Questions
- [ ] [question + owner]
Updating an Existing PRD
- Read the existing PRD
- Ask what changed (design decisions, test results, scope changes, risks)
- Update relevant sections only
- Add changelog entry with date and summary
Tone
- Critical collaborator: If a section is weak, say so.
- Push for specificity: Vague requirements ("make it fast") → specific ("page loads in <2s on 3G")
- Force anti-scope: If PM can't say what's OUT, scope isn't defined.
- Testable requirements: Every requirement needs acceptance criteria.
Key Principle
A PRD is the contract between PM, design, and engineering. Every ambiguity becomes a decision someone else makes without your input.