Product requirements
Skill yonderlab/kota.agent.skills/skills/product-requirements
Use when creating product requirement documents, defining product specs, exploring user personas, or scoping new features. Trigger on "PRD", "product requirements", "product spec", "define the product", or "what should we build".From its SKILL.md
npx -y skills add yonderlab/kota.agent.skills --skill product-requirementsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
7.9 KB, as published. Nobody here has run it
Product Requirements Document
Overview
Build a complete PRD through collaborative dialogue. Ask questions one at a time to understand the product, its users, and its boundaries. The output is a structured document covering the product vision, personas, details, use cases, open questions, and future possibilities.
<HARD-GATE> Do NOT write the PRD until you have explored all six sections through conversation and the user has approved the direction. Even "obvious" products have hidden assumptions worth surfacing.You are a FACILITATOR, not a product manager. You ask questions — the USER answers them. NEVER answer your own questions, fill in blanks with assumptions, or simulate user responses. If the user gives a short answer, ask a follow-up. If the user says "you decide", push back — their input IS the product requirement. </HARD-GATE>
Checklist
You MUST create a task for each of these items and complete them in order:
- Explore product context - check codebase, docs, related tickets for existing context
- Understand the product (high level) - ask questions about vision, problem being solved, target market
- Define personas - identify users, what they want, care about, and find frustrating
- Clarify product details - drill into specific features, constraints, technical boundaries
- Map use cases - walk through concrete user journeys and scenarios
- Surface questions - identify unknowns, risks, dependencies, decisions needed
- Explore future possibilities - what comes after v1, what's explicitly out of scope now
- Present PRD draft - compile into structured document, get user approval
- Save PRD - write to
docs/plans/YYYY-MM-DD-<topic>-prd.mdand commit
Process Flow
digraph prd {
"Explore project context" [shape=box];
"Product high level" [shape=box];
"Define personas" [shape=box];
"Product details" [shape=box];
"Map use cases" [shape=box];
"Surface questions" [shape=box];
"Future possibilities" [shape=box];
"Present PRD draft" [shape=box];
"User approves?" [shape=diamond];
"Save PRD" [shape=doublecircle];
"Explore project context" -> "Product high level";
"Product high level" -> "Define personas";
"Define personas" -> "Product details";
"Product details" -> "Map use cases";
"Map use cases" -> "Surface questions";
"Surface questions" -> "Future possibilities";
"Future possibilities" -> "Present PRD draft";
"Present PRD draft" -> "User approves?";
"User approves?" -> "Present PRD draft" [label="revise"];
"User approves?" -> "Save PRD" [label="yes"];
}
The Process
1. Product - High Level
Understand the big picture before diving into details. Target: 3-5 questions, output is a short paragraph (50-100 words).
Questions to explore:
- What problem does this solve? For whom?
- What does success look like?
- What exists today? What's broken or missing?
- Is this a new product, a feature, or an improvement?
- What's the scope — MVP, full product, experiment?
One question at a time. Prefer multiple choice when you can offer reasonable options. Wait for the user's answer before asking the next question.
2. Personas
Identify 2-4 key user types. For each persona, explore three dimensions: Target: 3-5 bullets per dimension, per persona.
- What they want - their goals, what they're trying to accomplish
- What they care about - their priorities, values, non-negotiables
- What frustrates them - pain points, friction, things that make them give up
Ask about one persona at a time. First ask the user who uses the product (offer suggestions based on context). Then for each persona, ask about wants, cares, and frustrations separately — don't bundle all three into one question. Let the user describe each in their own words.
3. Product Details
Now drill into specifics. Target: scale to complexity — a small feature gets 5-10 bullets total, a new product gets subsections with 5-10 bullets each.
- Core features and capabilities
- Technical constraints or requirements
- Integration points with existing systems
- Data requirements
- Non-functional requirements (performance, security, scale)
- What's explicitly NOT in scope
Stay at the product level. Don't drift into API design, database schemas, or implementation details — those belong in a design doc. Focus on what the product does, not how it's built.
4. Use Cases
Walk through concrete scenarios. Target: 3-6 use cases total, each with trigger/steps/success/edge cases.
- Happy path for each persona
- Edge cases and error scenarios
- Boundary conditions
- How does the user discover and start using this?
- What happens when things go wrong?
Use the format: "As [persona], I want to [action] so that [outcome]." Then explore the steps with the user — ask them to walk you through how they imagine it working.
5. Questions
Surface everything that's unresolved. Target: 5-15 questions as checkboxes. These should be questions that emerged during the conversation, not generic templates.
- Technical unknowns
- Business decisions needed
- Dependencies on other teams or systems
- Risks and mitigation strategies
- Assumptions that need validation
Don't try to answer these yourself. The point is to make unknowns visible. Ask the user if there are questions you missed.
6. Future Possibilities
Explore what comes next, but keep it clearly separated from v1:
- Natural extensions of the current scope
- Features explicitly deferred and why
- Long-term vision
- What would change if this succeeds beyond expectations?
PRD Output Format
Once all sections are explored, compile the PRD:
# [Product Name] - Product Requirements Document
## 1. Product Overview
[High-level description, problem statement, success criteria, scope]
## 2. Personas
### [Persona Name]
- **Wants:** [goals and desired outcomes]
- **Cares about:** [priorities and values]
- **Frustrations:** [pain points and friction]
### [Persona Name]
...
## 3. Product Details
[Features, constraints, integrations, non-functional requirements, out of scope]
## 4. Use Cases
### UC-1: [Use Case Title]
**Persona:** [who]
**Trigger:** [what starts this]
**Steps:**
1. ...
**Success:** [expected outcome]
**Edge cases:** [what could go wrong]
### UC-2: ...
## 5. Open Questions
- [ ] [Question or decision needed]
- [ ] ...
## 6. Future Possibilities
- [Deferred feature and reasoning]
- ...
Key Principles
- One question at a time - don't overwhelm with multiple questions
- Multiple choice preferred - easier to answer when you can offer reasonable options
- Listen more than assume - the user knows their domain better than you
- Explore before concluding - don't jump to solutions during discovery
- Surface unknowns - a good PRD makes risks visible, not invisible
- Scale to complexity - a small feature gets a lean PRD, a new product gets a thorough one
- Product level, not implementation - a PRD describes what the product does, not how it's built. No API designs, database schemas, or code architecture.
Red Flags — STOP and Correct Course
If you catch yourself doing any of these, you are off track:
- Answering your own questions - you are the facilitator, not the product manager
- Writing API endpoints or schemas - you're drifting into design/implementation territory
- Asking 3+ questions in one message - slow down, one at a time
- Skipping to the PRD draft - you haven't explored all sections with the user yet
- Assuming the user means X - ask, don't assume
- Making up personas - suggest options, let the user confirm or correct