Domain-Modeling — Keep the Language Sharp

The shared language decays silently: someone renames a concept, a term gains a second meaning, code and docs drift apart. This skill repairs it.

Audit

1. Read CONTEXT.md (if none, start from zero) + skim the codebase for repeated concepts
2. Compare: for every glossary term, does the code still use it? (grep the term — class names, function names, ACF field names, CSS classes)
3. Find drift: concepts in code with no term (grep a concept, several names come back), terms in CONTEXT.md no code references (dead vocabulary)

Challenge & stress-test

For each disputed term:

• Edge-case scenario: "two pages both call themselves the {term} — is that the same thing?" If the answer is "actually different", split the term
• Naming consistency: files, functions, ACF fields, and hooks should all use the term (or its prefix) — list any that don't

Update

• CONTEXT.md: add/split/remove terms with one-line definitions
• docs/adr/: a renamed concept or a split term is a decision worth recording
• Flag code that contradicts the model — rename only if the user approves (naming changes ripple; a cross-cutting rename is its own ticket)

Rules

• Terminology wins over code, but you never unilaterally rename production code — propose the rename, name the blast radius, let the user decide
• Keep the glossary terse: one line per term, no essays
• This is a deepening tool, not a churn tool — run it after substantial work, not after every commit