| name | prd |
| description | Write a real Product Requirements Document at the bar a senior/staff engineer or PM would ship at a top org (Google, Meta, Amazon, Apple). Gathers context, grills the user on the gaps, then produces a rigorous PRD: crisp problem statement, measurable goals & explicit non-goals, success metrics with baseline→target, prioritized requirements (P0/P1/P2), user journeys, risks, dependencies, and a phased rollout. Standalone — its output can feed /scope but it is NOT in the eng doc schema. Use when the user wants a PRD, product spec, one-pager, or PR/FAQ. Triggers: "write a PRD", "product spec", "product requirements", "spec this out", "one-pager", "PR/FAQ". |
prd — staff-level Product Requirements Document
Produce a PRD at the standard of a senior/staff engineer or PM at a top-tier org.
A great PRD is about the what and the why, not the how. It is crisp,
falsifiable, and decision-forcing — not a wall of prose. Ruthless about goals vs
non-goals, honest about risks, and every success metric is measurable.
This skill is independent of the eng plugin. Its output is a polished PRD
document; /scope can later read it to generate tickets, but you do NOT format
it in the eng doc schema.
Process
1. Gather context (don't write yet)
- Understand the idea from the user. Pull in any existing material (linked docs,
the codebase, prior art).
- If
enernet-personas (or a personas skill) is available and relevant, load it
so the PRD references real personas, not invented ones.
2. Grill the gaps before writing
A PRD is only as good as its inputs. Ask the user — in small batches, one topic
at a time, with your recommended answer for each — about whatever is unclear:
- Problem: whose problem, how painful, how do we know it's real (evidence)?
- Goal & success: what does winning look like, and what metric proves it?
- Scope: what's explicitly out? what's the smallest valuable version?
- Constraints: deadlines, platforms, compliance, dependencies.
Don't invent answers to material questions — get them, or list them as Open
Questions.
3. Choose the format
Default to the standard PRD below. Offer the Amazon PR/FAQ style (a mock
press release + anticipated FAQ) when the user is pitching a new product/feature
and wants to "work backwards." Ask only if it's ambiguous.
4. Write the PRD
Save to <project>/.plan/<feature-slug>/PRD.md by default (so /scope can find
it as a sibling), or wherever the user prefers (e.g. docs/prd/). Use this
structure:
# <Product / Feature name> — PRD
| Field | |
|-------|---|
| **Author** | <name> |
| **Status** | Draft / In Review / Approved |
| **Last updated** | <YYYY-MM-DD> |
| **Reviewers / Stakeholders** | <eng, design, PM, …> |
## TL;DR
<3-5 sentences a busy exec can read: what, for whom, why now, expected impact.>
## Background & Context
<What exists today, what changed, why this is worth doing now. Evidence/data.>
## Problem Statement
<The user/business problem in one sharp paragraph. Who hurts and how much.>
## Goals
<Outcome-oriented and measurable. Each goal ties to a success metric.>
- G1: …
## Non-Goals
<Explicitly out of scope. This section prevents creep and is as important as Goals.>
- NG1: …
## Success Metrics
<Each metric: name, baseline (today), target, how/when measured. No vanity metrics.>
| Metric | Baseline | Target | How measured |
|--------|----------|--------|--------------|
| … | … | … | … |
## Target Users & Personas
<Who this is for. Reference real personas. Their context, jobs-to-be-done.>
## User Journeys / Use Cases
<The key flows, before vs after. Walk the primary journey end to end.>
## Requirements
<Prioritized. P0 = must-have for launch, P1 = should, P2 = nice-to-have.>
### Functional
| ID | Priority | Requirement | Rationale |
|----|----------|-------------|-----------|
| FR1 | P0 | … | … |
### Non-Functional
<Performance, scale, security/privacy, accessibility, reliability, i18n.>
## Proposed Solution (high level)
<The what, not the detailed how. A sketch of the approach, key UX, system shape.
Leave detailed design to an eng design doc / the eng pipeline.>
## Risks & Mitigations
| Risk | Likelihood | Impact | Mitigation |
|------|-----------|--------|------------|
| … | … | … | … |
## Dependencies
<Teams, services, data, third parties this relies on.>
## Rollout / Phasing
<Milestones, phased delivery (MVP → v1 → v2), flagging, migration/backfill,
launch plan, how we de-risk.>
## Open Questions
<Honest list of unresolved decisions, owners, and needed-by dates.>
## Appendix / References
<Links to data, designs, prior docs, related ADRs.>
(If PR/FAQ format: lead with a one-page mock press release — headline, dated
dateline, customer quote, problem→solution narrative — followed by an
anticipated FAQ for customers and for internal stakeholders.)
5. Quality bar before you finish
- Goals are outcomes, not features; each has a metric.
- Non-goals are real and specific.
- Every success metric has a baseline and a target.
- Requirements are prioritized; P0 is genuinely minimal.
- Risks are honest (include the ones that could kill it).
- No solution detail masquerading as requirements.
6. Hand off
Tell the user the path and summarize. If they'll build it, suggest:
"Run /user-stories to break this into stories, or /scope to turn it into
eng tickets."