Development Using Claude Code or Codex · Part 3
Part 3: PRD Intake Skill
April 9, 2026 · 2 min read
PrdRequirementsPlanning
At this stage, you are not writing code. You are converting product intent into engineering clarity.
Skill for This Post: prd-intake
Create .ai/skills/prd-intake/SKILL.md:
---
name: prd-intake
description: Transform PRD/spec content into a clear engineering brief with assumptions, risks, and acceptance criteria.
---
# PRD Intake
## Checklist
- Capture problem statement in one paragraph
- Identify user personas and target outcomes
- Extract hard constraints (timeline, compliance, platform, performance)
- List ambiguous requirements as explicit open questions
- Define acceptance criteria in testable form
- Flag risks and dependencies
## Required Output
1) Scope: in-scope / out-of-scope
2) Assumptions list
3) Open questions list
4) Acceptance criteria checklist
5) Dependency mapPrompt Pattern
Run prd-intake on this PRD.
Output only: scope, assumptions, open questions, acceptance criteria, dependencies.
If anything is ambiguous, ask questions before task breakdown.Why This Skill Matters
- Prevents expensive rework caused by vague product language
- Creates a stable handoff for task breakdown
- Keeps Claude Code and Codex aligned on the same requirements
Cross-Agent Install
ln -sfn "$PWD/.ai/skills/prd-intake" ~/.claude/skills/prd-intake
ln -sfn "$PWD/.ai/skills/prd-intake" ~/.codex/skills/prd-intakeNext
Part 4 turns this intake artifact into dependency-aware implementation tasks with effort and risk tagging.