PRD Template Reference
How does our Tight Five PRD structure map to industry-standard PRD templates?
Our Template
We use a Tight Five structure: five sections that mirror the five questions every capability must answer. The operational template lives in the engineering skill:
- Active template:
.agents/skills/create-prd/references/prd-template.md - Process doc: AI Product Requirements
- Commissioning: L0-L4 Protocol
Every PRD produces four directories: pictures/ (thinking) → index.md (decision surface) → prompt-deck/ (sales) → spec/ (engineering).
Mapping
How traditional PRD sections translate to our Tight Five structure:
| Traditional PRD Section | Our Equivalent | Location |
|---|---|---|
| Executive Summary | Index (opening question + T-shape links) | prd-{name}/index.md |
| Problem Statement | Principles → The Job (SIO pattern) | spec/index.md#principles |
| User Stories / Personas | Players → Demand-Side Jobs (5 elements each) | spec/index.md#players |
| Functional Requirements | Build Contract (Feature/Function/Outcome table) | spec/index.md#build-contract |
| Technical Requirements | Platform → Current State + Build Ratio | spec/index.md#platform |
| Success Metrics | Performance → Quality Targets + Kill Signal | spec/index.md#performance |
| Implementation Plan | Protocols → Build Order (sprint sequence) | spec/index.md#protocols |
| Risk Assessment | Performance → Failure Budget + Kill Signal | spec/index.md#performance |
| Competitive Analysis | Principles → Design Constraints | spec/index.md#principles |
| Prioritization | Scorecard → 5P scoring (Pain x Demand x Edge x Trend x Conversion) | index.md#scorecard or spec/index.md#performance |
Key Differences
| Traditional | Tight Five |
|---|---|
| Acceptance criteria: pass/fail | Quality targets: percentage above threshold |
| MoSCoW prioritization | 5P composite score (product of five dimensions) |
| Milestones and timelines | Build order with sprint-sequenced feature refs |
| "Done" or "Not done" | Feature state enum (Gap → Stub → Built → Live) + L0-L4 maturity |
| Human-only document | Agent-facing spec (commands, boundaries, test contract) |
| Ship once | Commission in stages (Install → Test → Operational → Optimize) |
Industry References
- Lenny Rachitsky — PRD Templates — Problem before solution, one page, non-goals section
- Miqdad Jaffer (OpenAI) — AI PRD Template — AI-specific considerations, GTM
- Addy Osmani (Google) — Specs for AI Agents — Agent Experience (AX), modular context
- Jama Software — Writing Effective PRDs — Requirements management fundamentals
- Chisel Labs — PRD Templates — Competitive advantage framing
- Perforce — How to Write a PRD — MoSCoW prioritization
- Context Engineering — PRD Template — Living document principles
Context
- AI Product Requirements — The full Tight Five PRD methodology
- Commissioning Protocol — L0-L4 maturity model
- Standard Templates — Template inventory
Questions
When does a team need the traditional 9-section PRD structure — and when does the compressed Tight Five serve better?
- What do traditional PRDs capture that Tight Five misses — and is that a feature or a gap?
- If agent-facing specs become standard, does the traditional PRD format become purely historical?