Microcopy voice pack

A microcopy voice pack ships tone-consistent strings for the dozens of surfaces a build exposes: empty states, error messages, CTAs, tooltips, button labels, form validation. Voice packs are how a build sounds the same across widgets authored by different developers.

What you ship

my-voice/
├── manifest.json
├── strings/
│   ├── empty-states.json
│   ├── errors.json
│   ├── ctas.json
│   ├── tooltips.json
│   ├── validation.json
│   └── confirm-dialogs.json
├── voice-profile.json     # tonal markers (friendly, dry-witty, hospitality-warm, …)
├── preview/
│   └── samples.png        # 1200x630 sample-quote grid
└── README.md

Key contracts

• Keyed by intent, not by widget. The runtime resolves error.network.retry to your string regardless of which widget surfaced it.
• Voice profile up-front. Declare the register (friendly-pro, dry-witty, hospitality-warm, technical-precise) so the gate can spot tonal drift.
• i18n-ready. Strings target a single locale per pack; ship sibling packs for additional locales.

Scaffold

vincia create microcopy-voice-pack my-voice
cd my-voice
vincia preview
vincia publish --dry-run

Approval gates

1. Key coverage. At least the M-10 canonical key set must be present. 2. Voice profile compliance. A short LLM-free lint catches obvious mismatches (e.g. expletives in a hospitality profile). 3. No HTML. Strings are plain text; escape sequences are explicit.

Where to learn more

• /foundations — KitPreset.voice slot + the M-10 key catalog.