/nexus

Hallmark Recipe — Brand Identity Package Quality-Max

/nexus hallmark "<scope>" — Brand identity package quality-max — brand-core dialogue → identity tournament → persona-resonance + adversarial gauntlet → proof-carrying Brand Book + design tokens. The Quality-Max member that creates a product/org brand identity as a verified artifact: where growth-acceptance verifies brand tone at ship time (G14), hallmark builds the identity those checks verify against. Mostly no code (token definitions only).

Read this file before executing the hallmark Recipe.

1. Nature / When to Use / Boundaries

Hallmark is a proof-carrying brand construction recipe, not a document generator. Its identity comes from three things no sibling carries together:

1. Brand Core dialogue — a contract-level dialogue (spec-tier) excavates values, voice, personality, and explicit prohibitions before any visual work starts. An identity built on an unratified core is decoration, not brand.
2. Persona-resonance + differentiation gauntlet — the identity must survive (a) a persona panel that reacts to it as the real audience would, and (b) an adversarial refutation panel whose central attack is the logo-swap test: "does this identity still hold if a competitor's logo is swapped in?" An identity that survives the swap is interchangeable — refuted.
3. Proof-carrying Brand Book — the deliverable is a Brand Book + design tokens + application guide whose every load-bearing claim traces to a ratified dialogue decision or a gauntlet verdict, gated by the Doc Quality Gate.

Use hallmark for

• Building a product/org brand identity from scratch (naming direction, values, voice, visual identity, tokens).
• Overhauling an existing identity where the core is in question, not just the surfaces.
• Producing the brand source-of-truth that downstream recipes consume: rebrand (propagation), marquee (acquisition LP), growth-acceptance (Brand Compiler input).

Not this — route to

Scale: 10-24 agents, 3-8× cost. Mostly no code (token definitions only). Per-phase: P1 ≈ 3-4 · P2 ≈ 2 (dialogue-heavy, agent-light) · P3 ≈ 6-9 (3 directions × identity crafts) · P4 ≈ 5-8 (persona panel + refutation panel + Canon) · P5 ≈ 2-3.

Model selection (Plan-and-Execute): Phase 2 Brand Core dialogue and the Phase 4 refutation verdicts are judgment-heavy → plan-tier (opus / Fable 5). Phase 3 identity generators and Phase 4 persona panelists → Sonnet 5 default, per SKILL.md § Core Contract.

2. Termination Bound

The Phase 4 gauntlet is a loop ≤ N cycles (default N=2): each cycle re-crafts the identity against surviving attacks and re-runs the panels. Exit reasons use the canonical vocabulary:

On any non-ACCEPT exit the recipe reports best-so-far + the residual gap — never silently stops. Phases 0-3 and 5 are non-looping (N/A).

3. Confirm / Safety Gate

• Phase 0 Scope Gate: Confirm before launch — classify the ask as new-identity / full-rebrand-core / partial-refresh, surface the cost envelope and the downstream consumers (rebrand, marquee, growth-acceptance) so the user knows what this artifact feeds.
• Phase 2 Brand Core: contract-level dialogue; AUTORUN cannot skip (per reference/dialogue-protocol.md — same tier as spec dialogue). Values, voice, and prohibitions are user decisions, never inferred silently.
• Phase 5 sign-off: the Brand Book is presented for explicit ratification before it is written as the locked source of truth (it will govern every downstream surface).
• Standard Ask First tiers apply unchanged (hallmark rarely trips them — no destructive actions, minimal code).

4. Resume

Draft-resume (hallmark resume — spec-style: the dialogue draft + identity drafts are the state). State persists incrementally to docs/brand/<slug>.draft.md: Phase 1 findings → Discovery section; Phase 2 ratified core + Assumption Ledger delta at each checkpoint; Phase 3 the three direction candidates + tournament verdict; Phase 4 per-cycle surviving-attack ledger; Phase 5 promotes the draft to the locked docs/brand/<slug>.md + token files. A current-phase marker lets hallmark resume re-enter at the last checkpoint, summarizing decisions-so-far in 3-5 lines — never silently restarting from Discovery.

5. Phase Contract (AUTORUN chain template)

Phase 0  SCOPE GATE ★Confirm-before-launch
         Nexus[classify: new / rebrand-core / partial-refresh + cost + downstream consumers]
   ▼
Phase 1  DISCOVER ‖
         Compete[positioning + LLM brand visibility] ‖ Field/Voice[user perception — how the
         brand is actually heard today] ‖ Lens[existing asset inventory: logos, palettes,
         voice fragments, token files]
         → Discovery Digest (position, perception gaps, reusable assets)
   ▼
Phase 2  BRAND CORE ★contract-level dialogue (AUTORUN cannot skip; plan-tier model)
         Magi+Prose drive per reference/dialogue-protocol.md: values (3-5, ranked) · voice
         (register, vocabulary, tone anchors) · personality · explicit prohibitions
         ("never says / never looks like") · audience personas ratified
         → Brand Core Contract (every element elicited/ratified, Assumption Ledger for gaps)
   ▼
Phase 3  IDENTITY TOURNAMENT ‖ 3 directions from angles surfaced in the dialogue
         each: Vision[direction] → Muse[tokenize: color/type/spacing] + Ink[icon policy]
         + Builder[image]?[key visual] — token-first so every direction is application-ready
         → judge panel scores vs the Brand Core Contract → winner + salvage list
   ▼
Phase 4  GAUNTLET loop ≤ 2 cycles (default 2)
         ‖ Cast+Echo[demand]+Echo persona-resonance panel (each ratified persona reacts: trust,
           recall, emotional read — misresonance = attack)
         ‖ refutation panel per _common/ADVERSARIAL_REFUTATION.md — central attack is the
           logo-swap test: "swap in a competitor's logo; does the identity still hold?"
           surviving = interchangeable = refuted; plus core-contradiction and
           prohibition-violation attacks
         ‖ Canon[accessibility: palette contrast (WCAG), legibility at scale]
         exit per §2 (ACCEPT | diminishing-returns | cap-reached | BLOCK→Phase 2)
   ▼
Phase 5  COMPILE + DELIVER
         Prose+Scribe compile Brand Book (core, voice guide with do/don't pairs, visual
         identity, application guide) + Muse emits design tokens
         → Doc Quality Gate per reference/doc-quality-protocol.md (W1-W12; single source
           of truth — the Book may not contradict the tokens)
         → ★user ratification → lock docs/brand/<slug>.md + tokens
         → Hallmark Charter emitted; handoff recommendations (rebrand / marquee /
           growth-acceptance Brand Compiler) recorded, not executed

6. Output Report — Hallmark Charter (named)

Emitted inside NEXUS_COMPLETE on top of the base ## Nexus Execution Report:

• Brand Core Contract recap — values, voice, prohibitions, personas; each element's provenance (elicited / ratified / parked — none silent)
• Tournament verdict — 3 directions, per-direction scores vs the core, winner rationale + salvage applied
• Gauntlet ledger — persona-resonance results per persona; refutations raised / killed / survived (logo-swap verdict called out explicitly); Canon accessibility results
• Exit reason (§2 vocabulary) + residual gap (empty only on ACCEPT)
• Deliverable manifest — Brand Book path, token files, application guide
• Downstream handoffs — recommended next recipes (rebrand, marquee, growth-acceptance) with what each consumes

7. Failure Modes Prevented

8. Shared-Protocol References

9. Decision Tree vs Neighbors

Is the ask about a brand/identity (values, voice, look) rather than a feature/surface?
  NO  → the natural recipe for the task shape
  YES ↓
Does a ratified brand identity already exist, and the ask is applying it everywhere?
  YES → rebrand (propagation with completeness proof)
  NO ↓
Is it a person's professional brand (GitHub/LinkedIn/blog/conference)?
  YES → compete[brand]
  NO ↓
Is it one bounded design decision (a direction, a token set) with clear ownership?
  YES → vision / muse direct (minimum viable chain)
  NO ↓
Is the deliverable a conversion surface that consumes a brand (flagship LP)?
  YES → marquee (hallmark first if no Brand Book exists)
  NO  → hallmark — discover → core dialogue → tournament → gauntlet → Brand Book