prd-writer

Write enterprise-ready PRDs through guided interviews with ambiguity tracking and agent-executable epic plans.

Updated Apr 15, 2026
One-click install
npx skills add https://github.com/BING0116-gif/math-AI-assistant --skill prd-writer-bing0116-gif
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: prd-writer
Source: https://github.com/BING0116-gif/math-AI-assistant/tree/main/.agents/skills/prd-writer
Command: npx skills add https://github.com/BING0116-gif/math-AI-assistant --skill prd-writer-bing0116-gif

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Vague product requirements cause coding agents to drift, retry, and produce noisy pull requests. This Skill guides you through a structured interview to produce a PRD that works simultaneously as a product contract, an execution contract for coding agents, and a governance artifact. ## Core Features & Use Cases - Guided Section-by-Section Drafting: Walks through 17 PRD sections (problem statement, scope, functional requirements, tech stack, build/validate contract, security, observability, phase plan, epic cards) using interactive questions instead of assumptions. - Ambiguity Tracking: Logs every unresolved decision in a mandatory Unresolved Ambiguity Log appendix so nothing is silently guessed. - Agent-Ready Epic Sizing: Breaks work into session-sized epics with acceptance criteria, validation commands, and evidence requirements, using the included sizing rubric and full PRD template. - Use Case: A product manager needs a PRD for a new internal dashboard that a coding agent will implement. The Skill interviews them on scope, stack, and validation commands, then outputs a complete PRD.md with phased epics the agent can execute one session at a time. ## Quick Start Ask the assistant to run the prd-writer skill to draft a PRD for your feature, then answer its guided questions section by section.

Frequently Asked Questions about prd-writer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a PRD that a coding agent can execute?▼

Structure the PRD with explicit scope boundaries, an approved tech stack, exact build/test/lint commands, and epics sized for one agentic coding session. Each epic needs acceptance criteria, validation commands, and evidence requirements so the agent never guesses.

What sections should an enterprise PRD template include?▼

Include document control, problem statement, success metrics, scope and non-goals, functional and non-functional requirements, tech stack decision, architecture boundaries, build/validate contract, security and compliance, observability, phase plan, epic cards, and a definition of done.

How big should an epic be for an agentic coding session?▼

An epic should deliver one coherent outcome as a single reviewable PR with a small set of touched paths and finite validation commands. Avoid epics spanning multiple services, unresolved architecture decisions, or mixed discovery and implementation work.

How do I handle unknowns when drafting a PRD?▼

Record every unresolved item in an ambiguity log with the section, what is missing, its impact, and a suggested next step. Present the log before finalizing so stakeholders can resolve items or accept them as tracked open questions.

When should I not use a guided PRD writing process?▼

Skip the full process for trivial changes, spikes, or experiments where governance and agent execution contracts add no value. The structured approach pays off when work is handed to coding agents or requires security, compliance, and multi-reviewer approval.