📥 Consumers & promotion intake (start here)

[LinzLos]

Tiny Wire is vendored into several public repos. Those repos sometimes invent patterns that may belong upstream, or drift from current tokens. This issue is the two-way registry that keeps the coupling visible from the design-system side.

Consumer registry (Lane A — DS → consumer)

Reconcile each consumer when a Tiny Wire release ships:

• agentic-trust-ux — full lib/ copy (globals + components), byte-identical as of v1.5. Coupling map: Coupling map → Tiny Wire agentic-trust-ux#1
• dialing-prototype — tokens only (src/tokens/globals.css), diverged ~600 lines (uses --brand for success, dropped --ring/--text-disabled/elevation tier, hardcoded z-index). Coupling map: Coupling map → Tiny Wire dialing-prototype#1
• shift-prototype — tokens woven into src/index.css, adapted; TW component classes not used (rebuilt inline). Coupling map: Coupling map → Tiny Wire shift-prototype#1

Promotion candidates (Lane B — consumer → DS)

Patterns invented downstream, ranked by independent reinvention. Heuristic: reinvented in N consumers = promote.

• Segmented / tab control — reinvented in 3/3 consumers (agentic app.css:103, dialing App.css:709, shift QueueMonitor.tsx:127) → strongest case
• Icon system — shift (20+ inline SVGs), agentic (inline SVG); Tiny Wire has none today
• Live status pill + pulse dot — 3/3 wrote custom keyframes diverging from pulse-dot
• Trust Dial — agentic app.css:127
• Pre-flight decision surface — agentic app.css:161
• Attestation Shield — agentic index.html:84
• Search-with-clear input — shift Overview.tsx:295
• Cursor-tracking gradient card — shift Overview.tsx:491
• App-bar / top header — agentic .appbar (app.css:17); Tiny Wire has only a sidebar pattern, no top-nav component

Per-consumer candidate lists: LinzLos/agentic-trust-ux#2 · LinzLos/dialing-prototype#2 · LinzLos/shift-prototype#2

Assessment checklist (per candidate, when triaged)

Does it generalize beyond one prototype? · token/a11y debt · API shape · decision: promote / keep-local / reject

Open: consumption harmony (Lane A standardization)

Consumers currently vendor Tiny Wire three different ways — full lib/ copy (agentic), tokens-only file (dialing), tokens woven into app CSS (shift). A release therefore applies inconsistently, or silently not at all, per consumer. Converge on ONE consumption contract so every consumer picks up a release the same way.

• Decide the standard vendoring method (full lib/ copy vs. tokens-only import vs. published package)
• Migrate all three consumers to it; record each consumer's pinned Tiny Wire version

Relates to #1 (gated propagation), #2 (demo drift).

[LinzLos]

Consumption harmony — decision locked (2026-06-15)

Contract: full lib/ (tokens + components.css) is the single source of truth, vendored verbatim into every consumer. Prototypes consume the shared component classes (.card/.btn/.chip/.dot/.dialog) for standard chrome; bespoke/experimental UI stays local and promotion-tracked (Lane B). This makes a single edit in Tiny Wire flow to every prototype, and makes component drift structurally impossible (no hand-edits to vendored files).

Delivery: local sync script + version stamp + drift check. npm deferred (keeps the no-build ethos).

Migration sequence (low-risk):

1. Lock structure here: canonical vendored unit = lib/globals.css + lib/components.css with a /* Tiny Wire vX.Y */ stamp; document the app-override pattern (consumers never edit vendored files; overrides live in app-tokens.css/app.css).
2. agentic-trust-ux first (closest — already verbatim + uses classes).
3. shift-prototype — extract tokens from src/index.css into a clean vendored lib/; adopt component classes for chrome.
4. dialing-prototype (biggest lift) — unwind the ~600-line token fork; restore globals.css verbatim, move app-specific tokens to app-tokens.css; adopt component classes.
5. Build root sync script + per-repo drift check; extend the existing token-purity CI.
6. Record each consumer's pinned Tiny Wire version in the registry above once migrated.