ubiquitous-language

Installation
SKILL.md

What This Skill Does

Extract a DDD-style ubiquitous language glossary from a reviewed PRD and the codebase. Identify every domain term, flag ambiguities (same concept with different names, same name for different concepts), check how the codebase currently names things, and propose canonical terms with definitions.

This is the terminology alignment step in the planning pipeline (plan-featurewrite-a-prdreview-prdubiquitous-languagedesign-featurereview-plan). The PRD established what to build. This skill answers: "What do we call things — and does everyone agree?"

Why This Step Exists

Domain terminology drifts. The PRD says "workspace," the database calls it "org," the UI shows "team," and the API uses "account." When design-feature runs without a canonical vocabulary, it bakes naming inconsistencies into data models, API endpoints, and code constructs. Fixing terminology after design is expensive. Fixing it after implementation is worse.

This step creates a shared vocabulary that design-feature uses as its naming source of truth.

Finding the PRD

Look for the PRD in this order:

  1. If the user names a specific file, read that file.
  2. Check ./plans/ for PRD files (*-prd.md). If there's exactly one, use it. If there are several, ask which one.
  3. If the PRD was produced earlier in this conversation (via write-a-prd), use the conversation context.
Related skills

More from michaelmerrill/skills

Installs
1
First Seen
Mar 23, 2026