Skip to content
Back to skills

Generate Prd

ASecurity

Generates Product Requirements Documents for new features through guided discovery. Use when starting a new feature, creating specifications, or when user needs help defining requirements with clarifying questions.

  • 17 stars
  • 0 votes
  • 0 copies
  • 9 views
  • Added June 5, 2026
documentationgoperformance

Works with

  • cli

Security analysis

A100/100

Scanned June 5, 2026

npx -y skills add cfircoo/claude-code-toolkit --skill generate-prd --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Generate Prd?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Generate Prd
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/cfircoo-generate-prd/badge)](https://www.skillsdirectory.com/skills/cfircoo-generate-prd)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: generate-prd
description: Generates Product Requirements Documents for new features through guided discovery. Use when starting a new feature, creating specifications, or when user needs help defining requirements with clarifying questions.
model: claude-opus-4-5
context: fork
---

<objective>
Generate detailed, actionable Product Requirements Documents (PRDs) for new features. Guide users through clarifying questions, then produce specifications suitable for implementation by developers or AI agents.

**Important**: This skill creates PRDs only—it does NOT begin implementation.
</objective>

<quick_start>
1. Ask 3-5 clarifying questions with lettered options (A, B, C, D)
2. Generate structured PRD based on answers
3. Save to `tasks/prd-[feature-name].md`
</quick_start>

<process>

<step name="1_clarifying_questions">
Ask 3-5 essential questions using AskUserQuestion tool. Cover:

- **Problem/Goal**: What problem does this solve? Who benefits?
- **Core Functionality**: What are the must-have features?
- **Scope/Boundaries**: What is explicitly NOT included?
- **Success Criteria**: How will we know this worked?

Each question should have 3-4 concrete options (A, B, C, D format) plus "Other" for custom input.
</step>

<step name="2_generate_prd">
Create the PRD document with these sections:

1. **Introduction/Overview** – Brief feature description and problem statement
2. **Goals** – Specific, measurable objectives
3. **User Stories** – Each with title, description, and verifiable acceptance criteria
4. **Functional Requirements** – Numbered explicitly (FR-1, FR-2, etc.)
5. **Non-Goals** – Clear scope boundaries
6. **Design Considerations** – UI/UX requirements (if applicable)
7. **Technical Considerations** – Constraints and dependencies (if applicable)
8. **Success Metrics** – Measurable outcomes
9. **Open Questions** – Remaining clarifications needed
</step>

<step name="3_save_document">
Save the PRD to: `tasks/prd-[feature-name].md`

Use kebab-case for feature name (e.g., `prd-user-authentication.md`).
</step>

</process>

<principles>

<principle name="explicitness">
Avoid jargon or explain it. Provide concrete examples. A junior developer should understand every requirement without additional context.
</principle>

<principle name="verifiability">
Acceptance criteria must be testable and specific.

**Good**: "Login button redirects to /dashboard on success"
**Bad**: "Login works correctly"
</principle>

<principle name="ui_stories">
Every UI-focused user story MUST include this acceptance criterion:
`"Verify in browser using dev-browser skill"`
</principle>

<principle name="atomic_stories">
Each user story should be small enough to implement in one session. If a story feels too large, break it into smaller stories.
</principle>

</principles>

<prd_template>
```markdown
# PRD: {{Feature Name}}

## 1. Introduction

{{Brief description of the feature and the problem it solves}}

## 2. Goals

- {{Goal 1: Specific, measurable objective}}
- {{Goal 2: Specific, measurable objective}}

## 3. User Stories

### US-1: {{Story Title}}
**As a** {{user type}}, **I want** {{feature}} **so that** {{benefit}}.

**Acceptance Criteria:**
- {{Specific, testable criterion}}
- {{Specific, testable criterion}}
- Verify in browser using dev-browser skill (for UI stories)

### US-2: {{Story Title}}
...

## 4. Functional Requirements

- **FR-1**: {{Explicit requirement}}
- **FR-2**: {{Explicit requirement}}
- **FR-3**: {{Explicit requirement}}

## 5. Non-Goals

- {{What this feature explicitly does NOT do}}
- {{Scope boundary}}

## 6. Design Considerations

{{UI/UX requirements, wireframe references, design patterns}}

## 7. Technical Considerations

{{Constraints, dependencies, integration points, performance requirements}}

## 8. Success Metrics

- {{Measurable outcome 1}}
- {{Measurable outcome 2}}

## 9. Open Questions

- [ ] {{Unresolved question needing clarification}}
```
</prd_template>

<examples>

<example name="good_acceptance_criteria">
**Good:**
- "Search results display within 2 seconds"
- "Error message shows 'Invalid email format' for malformed input"
- "Clicking 'Save' disables button and shows spinner until complete"

**Bad:**
- "Search is fast"
- "Shows appropriate error messages"
- "Good UX on save"
</example>

<example name="good_functional_requirements">
- **FR-1**: System shall validate email format using RFC 5322 pattern
- **FR-2**: Password must be minimum 8 characters with at least one number
- **FR-3**: Session expires after 30 minutes of inactivity
</example>

</examples>

<success_criteria>
PRD is complete when:

- [ ] 3-5 clarifying questions were asked and answered
- [ ] All 9 PRD sections are filled with concrete content
- [ ] User stories have verifiable acceptance criteria
- [ ] UI stories include browser verification criterion
- [ ] Functional requirements are numbered (FR-1, FR-2, etc.)
- [ ] Non-goals clearly define scope boundaries
- [ ] Document saved to `tasks/prd-[feature-name].md`
- [ ] No implementation has begun (PRD only)
</success_criteria>

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.