name: prd-generator description: Standardized PRD generation skill for PM review agents. Generates consistent, wave-based PRDs with proper subagent assignments and quality criteria.

PRD Generator Skill

This skill provides standardized PRD (Product Requirements Document) generation for all PM review items. Every improvement submitted to the backlog MUST have a complete PRD.

PRD Structure

markdown

# PRD: [Title]

## Problem Statement

[Expanded from item.problem with user impact and business context]

## Proposed Solution

[Expanded from item.solution with specific implementation approach]

## Success Criteria

- [ ] [Measurable criterion 1]
- [ ] [Measurable criterion 2]
- [ ] [Measurable criterion 3]

## Technical Approach

### Wave 1: Foundation (Explore)

| #   | Subagent | Task                    |
| --- | -------- | ----------------------- |
| 1.1 | Explore  | [Research task]         |
| 1.2 | Explore  | [Identify dependencies] |

### Wave 2: Implementation (general-purpose)

| #   | Subagent        | Task                  |
| --- | --------------- | --------------------- |
| 2.1 | general-purpose | [Core implementation] |
| 2.2 | general-purpose | [Tests]               |

### Wave 3: Validation (code-reviewer)

| #   | Subagent      | Task             |
| --- | ------------- | ---------------- |
| 3.1 | code-reviewer | [Review changes] |

## Risks & Mitigations

| Risk                       | Mitigation |
| -------------------------- | ---------- |
| [Risk based on complexity] | [Strategy] |

## Out of Scope

- [Explicit exclusions]

Wave Structure Guidelines

Wave 1: Foundation (Always Required)

Purpose: Understand the codebase before making changes

Subagent: Explore

Tasks:

•Research existing patterns and implementations
•Identify affected files and dependencies
•Check for related existing code
•Understand data flow

Example tasks:

code

1.1 | Explore | Find existing auth patterns in src/lib/auth/
1.2 | Explore | Identify all components using the affected API
1.3 | Explore | Check test coverage for affected files

Wave 2: Implementation (Core Work)

Purpose: Make the actual changes

Subagent: general-purpose

Tasks:

•Implement core functionality
•Write tests
•Update documentation if needed

Example tasks:

code

2.1 | general-purpose | Implement auth check in route handler
2.2 | general-purpose | Add unit tests for new auth logic
2.3 | general-purpose | Update error messages for auth failures

Wave 3: Validation (Quality Gate)

Purpose: Verify changes meet standards

Subagent: code-reviewer

Tasks:

•Review all changes for quality
•Check for security issues
•Verify test coverage
•Ensure consistency with patterns

Example tasks:

code

3.1 | code-reviewer | Review all changes against project standards

Complexity-Based Wave Scaling

Scale the PRD complexity based on the item's complexity score (1-5):

Complexity	Waves	Tasks per Wave	Risk Section
1 (Trivial)	2	1-2	Optional
2 (Low)	2	2-3	1 risk
3 (Medium)	3	2-3	2 risks
4 (High)	3	3-4	3 risks
5 (Very High)	4	3-5	4+ risks

Complexity 1-2: Simple Changes

markdown

### Wave 1: Implementation (general-purpose)

| #   | Subagent        | Task         |
| --- | --------------- | ------------ |
| 1.1 | general-purpose | [Simple fix] |
| 1.2 | general-purpose | [Add test]   |

### Wave 2: Validation (code-reviewer)

| #   | Subagent      | Task         |
| --- | ------------- | ------------ |
| 2.1 | code-reviewer | Quick review |

Complexity 3: Medium Changes

markdown

### Wave 1: Foundation (Explore)

| #   | Subagent | Task                         |
| --- | -------- | ---------------------------- |
| 1.1 | Explore  | Research existing patterns   |
| 1.2 | Explore  | Identify affected components |

### Wave 2: Implementation (general-purpose)

| #   | Subagent        | Task                    |
| --- | --------------- | ----------------------- |
| 2.1 | general-purpose | Implement core change   |
| 2.2 | general-purpose | Write tests             |
| 2.3 | general-purpose | Update affected callers |

### Wave 3: Validation (code-reviewer)

| #   | Subagent      | Task                       |
| --- | ------------- | -------------------------- |
| 3.1 | code-reviewer | Full review of all changes |

Complexity 4-5: Major Changes

Add additional waves for:

•Database changes: Add migration wave before implementation
•API changes: Add contract verification wave
•UI changes: Add design review wave
•Security changes: Add security audit wave

markdown

### Wave 1: Foundation (Explore)

...

### Wave 2: Migration (general-purpose)

| #   | Subagent        | Task                    |
| --- | --------------- | ----------------------- |
| 2.1 | general-purpose | Create migration file   |
| 2.2 | general-purpose | Test migration rollback |

### Wave 3: Implementation (general-purpose)

...

### Wave 4: Validation (code-reviewer)

...

Success Criteria Guidelines

MUST be measurable and verifiable:

Good Success Criteria:

markdown

- [ ] API endpoint returns 403 for unauthorized users
- [ ] Unit tests pass with 100% coverage of new code
- [ ] Page load time under 200ms on cached requests
- [ ] No TypeScript errors or warnings

Bad Success Criteria (too vague):

markdown

- [ ] Code works correctly (HOW do we verify?)
- [ ] Performance is improved (BY HOW MUCH?)
- [ ] Users are happy (NOT MEASURABLE)

Risk Assessment Guidelines

Risk Categories:

Category	Example Risks
Data	Data loss, corruption, migration failure
Performance	Degraded response time, increased load
Security	New attack surface, auth bypass
Compatibility	Breaking changes, API contract changes
Dependencies	External service outage, version conflicts

Mitigation Strategies:

Risk Type	Common Mitigations
Data loss	Backup before migration, rollback plan
Performance	Load testing, feature flag, gradual rollout
Security	Security review, penetration testing
Compatibility	API versioning, deprecation period
Dependencies	Fallback service, circuit breaker

Out of Scope Guidelines

Always include explicit exclusions to prevent scope creep:

What to Exclude:

•Related but separate features - "This PRD does not cover user notifications"
•Performance optimizations - "Database indexing is separate work"
•Design changes - "Visual redesign is out of scope"
•Documentation - "API docs update is tracked separately"

Example:

markdown

## Out of Scope

- User notification system (tracked in separate PRD)
- Mobile responsive design for this component
- Integration with third-party analytics
- Internationalization of error messages

Category-Specific PRD Templates

Security PRD

Add after Success Criteria:

markdown

## Security Considerations

- **Threat model:** [What attacks does this prevent?]
- **Attack surface:** [Does this expand or reduce attack surface?]
- **Auth impact:** [How does this affect authentication/authorization?]

Performance PRD

Add after Success Criteria:

markdown

## Performance Targets

- **Baseline:** [Current metric]
- **Target:** [Expected improvement]
- **Measurement:** [How to verify]

Data PRD

Add after Success Criteria:

markdown

## Data Migration Plan

- **Backup:** [Backup strategy before migration]
- **Rollback:** [How to revert if needed]
- **Validation:** [How to verify data integrity]

PRD Generation Process

•Parse item fields: title, problem, solution, complexity, type
•Determine wave count: Based on complexity score
•Generate Problem Statement: Expand item.problem with context
•Generate Solution: Expand item.solution with implementation details
•Create Success Criteria: 3-5 measurable criteria
•Build Wave Structure: Using complexity scaling rules
•Add Risks: Based on complexity and type
•Define Out of Scope: Explicit exclusions
•Apply Category Template: If security/performance/data

Output Format

Return PRD as markdown string in prd_content field:

json

{
  "prd_content": "# PRD: [Title]\n\n## Problem Statement\n...",
  "prd_generated_at": "2026-01-31T12:00:00Z",
  "prd_version": "1.0",
  "complexity_used": 3,
  "waves_generated": 3
}