| name | create-prd |
| description | Create a PRD through user interview, codebase exploration, and module design, then submit as a GitHub issue, Azure DevOps work item, or local file. Use when the user wants to create or write a PRD, create a product requirements document, design a new feature, or capture requirements. |
Create PRD
Guide the user from a problem statement to a fully-formed PRD, then submit it to their preferred destination.
Skill workflow — these skills chain naturally:
create-prd → pre-mortem (optional stress-test) → plan-from-prd (phased plan) → prd-slice (tracker work items)
Process
1. Gather the problem statement
Ask the user for a detailed description of:
- The problem they are trying to solve
- Any ideas or constraints they already have in mind
2. Explore the codebase
Use a subagent to explore the repo and verify the user's assertions. Understand the current architecture, existing patterns, and potential integration points.
3. Interview the user
Relentlessly interview the user about every aspect of the plan until you reach a shared understanding. Walk down each branch of the design tree, resolving dependencies between decisions one-by-one. Ask one question at a time, and confirm understanding after each answer and always suggest the best answer based on your knowledge of the codebase and best practices. This is a recursive process that will continue until you have a clear, shared vision for the feature.
As user stories accumulate, continuously assess the overall complexity of the feature. When the scope feels large — many moving parts, cross-cutting concerns, or a broad user-facing surface area — challenge the user to identify a first sufficient MVP wave before expanding further. Ask: "What is the smallest version of this that delivers real value?" This is not about counting stories; it is about understanding whether the work can be meaningfully shipped in phases.
3a. Pre-mortem stress-test (optional)
If the PRD is growing large, assumptions are unclear, or requirements conflict, pause here and invoke the pre-mortem skill before proceeding to Step 4. It will interrogate the design, surface contradictions, and return either a risk document or an annotated PRD. Resume from Step 4 once pre-mortem is complete.
4. Design modules (optional — skip if the user wants a lightweight PRD)
Sketch the major modules to build or modify. Actively look for opportunities to extract deep modules — ones that encapsulate significant functionality behind a simple, stable, testable interface.
Review with the user:
- Do these modules match their expectations?
- Which modules should have tests written for them?
5. Write the PRD
Use the <prd-template> below to write the PRD.
6. Submit the PRD
Ask the user where they want it saved: GitHub, Azure DevOps, or local file.
Run a subagent to submit so you keep your main context clean:
- GitHub: Use the
gh-cli skill (gh issue create) with appropriate labels and assignees.
- Azure DevOps: Use the
azure-devops-cli skill (az boards work-item create) with appropriate tags and assignees.
- Local file: Save to
./prds/<feature-name>.md.
Problem Statement
The problem that the user is facing, from the user's perspective.
Solution, Done looks like, and User Experience
The solution to the problem, from the user's perspective. What 'done' looks like and the expected user experience.
User Stories
A LONG, numbered list of user stories. Each user story should be in the format of:
- As an , I want a , so that
1. As a mobile bank customer, I want to see balance on my accounts, so that I can make better informed decisions about my spending
This list of user stories should be extremely extensive and cover all aspects of the feature.
Implementation Waves (omit for small features)
When the feature is complex — many stories, broad surface area, or multiple cross-cutting concerns — describe how delivery will be broken into successive waves. Each wave should be independently demonstrable and deliver a coherent user experience on its own.
- Wave 1 (MVP): The smallest version that delivers real value. Define what "done" looks like for this wave.
- Wave 2: Additional depth or breadth that builds on Wave 1.
- (add waves as needed)
Prefer grouping stories into waves so that parallel workstreams within a wave share no blocking dependencies.
Implementation Decisions
A list of implementation decisions that were made. This can include:
- The modules that will be built/modified
- The interfaces of those modules that will be modified
- Technical clarifications from the developer
- Architectural decisions
- Schema changes
- API contracts
- Specific interactions
Do NOT include specific file paths or code snippets. They may end up being outdated very quickly.
Testing Decisions
A list of testing decisions that were made. Include:
- A description of what makes a good test (only test external behavior, not implementation details)
- Which modules will be tested
- Prior art for the tests (i.e. similar types of tests in the codebase)
Validation Criteria
A list of criteria that can be used to validate whether the implementation meets the requirements. This can include:
- Specific scenarios that must work
- Performance benchmarks
- User acceptance criteria
- Metrics to be collected
- Success criteria for each implementation wave
Out of Scope
A description of the things that are out of scope for this PRD.
Further Notes
Any further notes about the feature.
The skill's work is complete when the PRD has been submitted to the user's chosen destination. Return control to the user.