lintMarkup() is the dotgui linter — idiom, best-practice, and content checks beyond schema legality, the single source of truth for “is this good, idiomatic dotgui?”.

/lint

qualityidiom & content checks

The dotgui linter — idiom, best-practice, and content checks on top of schema legality. Where /validate answers “is this legal?”, lint answers “is this good, idiomatic dotgui?”.

It is the single source of truth for idiom rules, replacing the old per-consumer copies. Deterministic fixes for the issues it finds live in /autofix.

Signature

lintMarkup(xml: string): LintResult

interface LintResult {
  issues: LintIssue[]   // all findings
  errors: LintIssue[]   // level === 'error'
  ok: boolean           // no errors
  ran: boolean          // false only if markup couldn't be parsed
}

What it checks

Structure — unknown tags, empty spacers, required w/h, fill children needing a sized parent.
Attributes — invented attrs (margin, py, flex…), boolean="false", text-align on text, legacy stroke.
Values — color formats (hex/$token only), gap="auto" without a fill dimension, undefined $token refs.
Content “AI tells” — em/en-dashes, lorem ipsum, placeholder names/emails, emoji in copy, filler verbs.

Use cases

gui lint in the CLI.
gui-app authoring diagnostics.
The skill linter — one rule set, not a re-port.

Lint markup and report errors

import { lintMarkup } from '@dotgui/kit/lint'

const { ok, errors, issues } = lintMarkup(markup)
if (!ok) for (const e of errors) console.error(`[${e.where}] ${e.message}`)