Designly

View saved version →

---
name: arabic-rtl-director
description: Arabic visual hierarchy, RTL reading flow, calligraphy glyph fidelity, and regional cultural authenticity director. This skill should be used when designing Arabic-first posters and campaigns, directing Right-to-Left eye paths, balancing mixed Arabic/English typography, or issuing hard vetoes on Arabic glyph and rendering errors.
---

# Arabic RTL Director

Arabic RTL Director enforces native Arabic visual architecture, natural Right-to-Left (RTL) reading hierarchy, calligraphy glyph correctness, and regional cultural authenticity across the MENA region.

---

## 1. Core Workflow

1. **Native RTL Visual Flow**:
   - Establish the primary visual path: **Top-Right (Headline / Entry)** ➔ **Visual Center / Dynamic Diagonal** ➔ **Bottom-Left (CTA / Signoff / Secondary Mark)**.
   - **No Naive Mirroring**: Never perform mechanical horizontal flipping of an English layout. Rebuild spatial balance natively around Arabic typographic mass and focal anchors.

2. **Exact-Copy & Glyph Protection (Hard Gate)**:
   - Arabic text must be rendered with correct contextual glyph shaping (isolated, initial, medial, final forms).
   - Reject left-to-right character rendering, unjoined letter errors, or corrupt Unicode reversals.
   - Maintain exact client diacritics (Harakat) and preserve original copy wording without paraphrasing.

3. **Mixed Arabic/English Typographic Balance**:
   - When Latin brand names coexist with Arabic headlines, align optical x-heights and baseline weights.
   - Prevent the Latin word from visually overpowering or breaking the Arabic grammatical phrase.

4. **Cultural Resonance & Visual Archetypes**:
   - Account for regional visual nuances:
     - **GCC / Gulf**: Clean luxury, subtle geometry, refined warm desert tones, high-finish metallics.
     - **Egypt / Levant**: Vibrant cultural energy, expressive editorial typography, rich storytelling textures.
     - **North Africa**: Geometric architectural motifs, Mediterranean light, modern bilingual identity.

5. **Hard Veto Authority**:
   - Emit an immediate `hard_veto` if generated or proposed artwork features broken Arabic letterforms, reversed text, or inappropriate cultural symbolism.

6. **Output Contract**:
   - Return structured `typography_state.arabic_rtl` and `composition_state.rtl_flow` inside `DesignSignalPacket`.

---

## 2. Cross-Skill Neural Connections & References

### Peer & Specialist Skills
- [Typography Director](../typography-director/SKILL.md) — Font pairing, typographic measure, and line spacing
- [Composition Director](../composition-director/SKILL.md) — Grid architecture, focal balance, and RTL margins
- [Visual QA](../visual-qa/SKILL.md) — Arabic hard release gate & glyph verification
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and lock manager

### Schemas & References
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Signal handoff contract
- [Arabic RTL & Cultural Guide](../../shared/references/arabic-rtl-and-cultural.md) — Regional design rules
- [Typography Guide](../../shared/references/typography.md) — Typographic hierarchy
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Cultural authenticity standards

Referenced files: 3

View saved version →

---
name: brand-activation
description: Specialist in experiential campaigns, PR stunts, and brand utility. This skill should be used when creating non-traditional marketing activations, interactive physical/digital stunts, brand utility tools, cultural objects, ambient media, or applying the diagnostic test to distinguish genuine non-advertising utility from channel-specific executions.
---

# Brand Activation

Brand Activation designs physical, digital, and cultural acts that generate organic attention, cultural conversation, and real human utility beyond standard paid media.

---

## 1. The 9 Activation Formats

1. **Brand Utility**: Building a functional service, physical tool, or digital platform that solves an unaddressed consumer problem.
2. **PR Stunt / Guerilla**: An unexpected public demonstration engineered to trigger news coverage and social commentary.
3. **Cultural Hijack**: Repurposing an established cultural moment, landmark, or public ritual.
4. **Subversive Demonstration**: A provocative physical torture test or comparative experiment in public space.
5. **Interactive Installation / Ambient**: Immersive, multi-sensory physical environments inviting direct participant manipulation.
6. **Social Experiment**: Staging an authentic human situation in public to reveal and challenge social norms.
7. **Product Innovation / Packaging Hack**: Re-engineering the physical product or packaging to deliver a political or environmental statement.
8. **Digital-Physical Bridge**: Connecting physical objects to real-time generative code, AR, or digital ecosystems.
9. **Cultural Object / Artefact**: Creating an enduring, tangible artefact that outlives the marketing cycle.

---

## 2. The Non-Advertising Diagnostic Test

Before approving an activation concept, apply the diagnostic test:

> **"If we remove the advertising campaign entirely, does this object or act still possess intrinsic value and cultural meaning?"**
> - **Yes** $\rightarrow$ `non_advertising` platform (Scalable, high earned media, award-grade).
> - **No** $\rightarrow$ `execution` mechanic (Single-format tactic, dependent on media spend).

---

## 3. Activation Decision Matrix

- **High Budget / High Risk**: Physical utility or cultural infrastructure (e.g. permanent public installations).
- **Low Budget / High Originality**: Cultural hijack or packaging subversion.
- **Participation Floor**: The physical mechanic must be intuitively understandable within 3 seconds of contact.

---

## 4. Cross-Skill Neural Connections & References

### Peer & Specialized Skills
- [Campaign Canon](../campaign-canon/SKILL.md) — Canonical activation case benchmarks & P04/P07 patterns
- [Creative Director](../creative-director/SKILL.md) — Cannes scoring & recursive ideation
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Brand guidelines, purpose & brand-off test
- [Manipulation Director](../manipulation-director/SKILL.md) — Real-world physical plausibility and lighting
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator

### Domain References & Contracts
- [Activation Toolkit](references/activation-toolkit.md) — Comprehensive format matrix and mechanics
- [Idea Taxonomy](references/idea-taxonomy.md) — 7-level taxonomy and activation diagnostics
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-bias & ethical standards
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh signal handoff
- [Design Context Schema](../../shared/contracts/design-context.schema.json) — Shared state context

Referenced files: 6

View saved version →

---
name: brand-intelligence
description: Brand identity and product fidelity specialist. This skill should be used when auditing visual proposals against brand guidelines, enforcing logo clearspace and color formulas, verifying product packaging fidelity, conducting the Brand-Off test, or issuing hard vetoes on brand violations.
---

# Brand Intelligence

Brand Intelligence enforces brand integrity, logo clearspace, color formulas, and physical product fidelity across all creative output. It acts as the guardian of documented brand guidelines and has hard-veto authority over conflicting taste or stylistic proposals.

---

## 1. Core Workflow

1. **Brand Asset Ingestion**:
   - Parse official brand manuals, hex codes, typefaces, and vector logo files.
   - Extract mandatory product packaging geometry, label placements, and unalterable elements.

2. **Rule Level Distinction**:
   - **Documented Brand Rules (Priority 2, Confidence 1.00)**: Immutable constraints (e.g. Pantone / Hex codes, minimum logo clearspace, forbidden background colors). Cannot be overridden by taste preferences.
   - **Observed Brand Patterns (Priority 10, Confidence 0.75-0.90)**: Inferred brand tendencies from past campaigns. Can be adapted for specific campaign objectives.

3. **The Brand-Off Test**:
   - Mental or visual test: If we swap out the brand logo and replace it with a direct competitor's logo, does the visual still make sense?
   - If YES: The visual lacks brand specificity and is merely generic category imagery. Rework composition, palette, and materials to reflect unique brand DNA.

4. **Veto Issuance**:
   - If a proposed visual violates brand clearspace, alters logo proportions, uses forbidden gradient treatments on trademarked assets, or misrepresents product physical packaging: emit an immediate `hard_veto`.

5. **Output Contract**:
   - Return a valid `DesignSignalPacket` containing `brand_state` decisions, clearspace rules, and any active vetoes or warnings.

---

## 2. Hard Gates & Verification

- **Logo Protection**: Never place text or high-contrast busy imagery inside the defined logo exclusion zone (minimum 0.5x logo height).
- **Product Realism**: Physical dimensions, label text, and material finishes of real products must match reality with >= 98% fidelity.
- **Color Fidelity**: Brand core colors must match specified hex / CMYK / Pantone values without arbitrary color grading shifts.

---

## 3. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Creative Strategy](../creative-strategy/SKILL.md) — Strategic territory alignment & brand mission
- [Taste Engine](../taste-engine/SKILL.md) — Reconciling brand locks with reference aesthetics
- [Typography Director](../typography-director/SKILL.md) — Corporate typeface rules and font hierarchy
- [Campaign DNA](../campaign-dna/SKILL.md) — Multi-asset brand continuity across channels
- [Visual QA](../visual-qa/SKILL.md) — Brand fidelity floor & logo gate verification
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and lock enforcement

### Schemas & References
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh handoff
- [Brand Intelligence Guide](../../shared/references/brand-intelligence.md) — Brand analysis rules
- [Design Principles](../../shared/references/design-principles.md) — Core principles
- [Software Architecture](../../shared/references/software-architecture-and-contracts.md) — Priority hierarchy

Referenced files: 3

View saved version →

---
name: campaign-canon
description: Advertising history and campaign pattern benchmarking specialist. This skill should be used when referencing the 571 canonical advertising campaigns, looking up case studies across P01-P18 patterns, checking market pattern saturation, evaluating anti-derivative uniqueness, or finding cross-industry analog cases by industry, budget, format, or emotional intent.
---

# Campaign Canon

Campaign Canon connects modern commercial design to the world's most rigorous library of 571 verified legendary campaigns across Cannes, D&AD, and Effie history. It serves as an anti-derivative benchmark to ensure new work is strategically grounded, structurally distinct, and not merely repeating saturated clichés.

---

## 1. Pattern Taxonomy (P01–P18)

Every canonical campaign is categorized by its primary mechanic:

| Pattern | Name | Core Mechanic | Saturation Risk |
|---|---|---|---|
| `P01` | Extreme Consequence | Exaggerating the benefit to an absurd extreme | Low |
| `P02` | Inversion / Anti-Hero | Turning weakness or villain into the core proof | Medium |
| `P03` | Uncomfortable Truth | Stating the taboo or unsaid category reality | Medium |
| `P04` | Brand Utility | Solving a real human problem instead of just talking | Low |
| `P05` | Provocative Question | Re-framing common cultural assumptions | Medium |
| `P06` | Radical Transparency | Showing flaws, raw process, or internal secrets | Low |
| `P07` | Cultural Hijack | Intercepting an existing cultural event or conversation | High |
| `P08` | Product Torture Test | Pushing physical product capability to destruction | Medium |
| `P09` | Emotional Parable | Character-driven narrative demonstrating value | **Saturated (50+ cases)** |
| `P10` | Interactive / Gamified | Requiring active audience participation or code | Low |
| `P11` | Social Experiment | Staging real-world reactions to reveal human behavior | **Saturated (50+ cases)** |
| `P12` | Subversive Demonstration | Challenging competitor conventions via real demo | Medium |
| `P13` | Relatable Slice-of-Life | Capturing micro-moments of authentic daily struggle | Medium |
| `P14` | Visual Pun / Metaphor | Single-image visual synthesis delivering the claim | Medium |
| `P15` | Absurdist Comedy | Surrealistic non-sequitur logic anchoring memorability | Medium |
| `P16` | Cause / Purpose | Aligning brand power to a societal transformation | **Saturated (50+ cases)** |
| `P17` | Technology Demo | Harnessing novel tech/AI to dramatize a benefit | Low |
| `P18` | Nostalgia / Cultural Remix | Re-contextualizing historical cultural icons | Medium |

---

## 2. Anti-Derivative & Saturation Rules

1. **Empirical Saturation Cap**:
   - If a proposed concept matches a saturated pattern (`P09`, `P11`, `P16`) with 50+ canonical cases, cap originality score at **6.0 / 10** unless a structurally distinct variant mechanic is introduced.
2. **Cross-Category Borrowing**:
   - Recombining proven mechanics from distant categories is encouraged and creates high-originality work.
3. **Pre-Mortem Calibration**:
   - Before signing off on a concept, compare it against 3–5 canonical cases in the same pattern to identify failure modes before launch.

---

## 3. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Creative Director](../creative-director/SKILL.md) — Cannes scoring & recursive ideation
- [Brand Activation](../brand-activation/SKILL.md) — PR stunts, ambient & brand utility mechanics
- [Creative Strategy](../creative-strategy/SKILL.md) — Strategic territory & message hierarchy
- [Taste Engine](../taste-engine/SKILL.md) — Visual reference deconstruction & rule extraction
- [Visual QA](../visual-qa/SKILL.md) — Hard gates & anti-slop release evaluation

### MOC Navigation Matrix & References
- [Master Index (MOC-index)](references/legendary-campaigns/MOC-index.md) — Complete case library navigation
- [Pattern Index (MOC-pattern)](references/legendary-campaigns/MOC-pattern.md) — Lookup by P01–P18 mechanics
- [Industry Index (MOC-industry)](references/legendary-campaigns/MOC-industry.md) — Lookup by market sector
- [Emotion Index (MOC-emotion)](references/legendary-campaigns/MOC-emotion.md) — Lookup by emotional tone
- [Budget Index (MOC-budget)](references/legendary-campaigns/MOC-budget.md) — Lookup by resource tier
- [Format Index (MOC-format)](references/legendary-campaigns/MOC-format.md) — Lookup by media channel
- [Tag Schema Contract](references/tag-schema.md) — 17-axis metadata schema
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh signal handoff
- [Design Context Schema](../../shared/contracts/design-context.schema.json) — Shared state context

Referenced files: 588

View saved version →

---
name: campaign-dna
description: Multi-asset campaign visual DNA and cross-format continuity director. This skill should be used when planning multi-asset campaigns, defining visual-family continuity rules across formats (1:1, 9:16, 16:9), creating deliberate variation across deliverables, or maintaining campaign brand consistency.
---

# Campaign DNA

Campaign DNA defines the visual genome for multi-asset commercial campaigns. It establishes unbreakable family resemblance across disparate formats and channels (social, digital OOH, print, performance ads) while introducing deliberate creative variation across individual assets.

---

## 1. Core Workflow

1. **Campaign Genome Definition**:
   - **Primary Palette DNA**: Core 2-3 colors that appear consistently in every asset.
   - **Lighting Signature**: Consistent light direction, contrast ratio, and color temperature across all campaign shots.
   - **Material & Texture DNA**: Common tactile finish (e.g. brushed metal, organic matte paper, high-gloss glass).
   - **Typographic Anchor**: Shared font pairing, tracking rules, and badge placement.

2. **Deliberate Variation Matrix**:
   - Never copy-paste the exact same layout across 5 assets.
   - Vary intentional dimensions while holding the genome locked:
     - Asset 1 (Hero / Brand): Macro product hero shot with dynamic dramatic lighting.
     - Asset 2 (Context / Lifestyle): Environmental wide shot showing product in real use.
     - Asset 3 (Detail / Feature): Close-up crop emphasizing tactile craft and engineering.
     - Asset 4 (Story / Conversion 9:16): Vertical composition with clear CTA zone.

3. **Multi-Format Responsive Adaptation**:
   - **1:1 Square (Feed)**: Centered or Rule-of-Thirds balance with equal margin breathing room.
   - **4:5 Portrait (Instagram Feed)**: Extended vertical hierarchy with upper headline and lower CTA.
   - **9:16 Vertical (Stories / TikTok)**: Top 15% and bottom 20% safe zone clearance for UI overlays.
   - **16:9 Landscape (YouTube / Desktop / DOOH)**: Dynamic horizontal eye path with distinct left/right narrative balance.

4. **Output Contract**:
   - Return structured `campaign_state` and `VisualDNA` JSON inside `DesignSignalPacket`.

---

## 2. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Creative Strategy](../creative-strategy/SKILL.md) — Campaign objectives, message hierarchy & audience
- [Visual Storytelling](../visual-storytelling/SKILL.md) — Multi-frame narrative arc continuity
- [Composition Director](../composition-director/SKILL.md) — Aspect ratio adaptations and safe zones
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Brand guidelines and multi-asset consistency
- [Prompt Compiler](../prompt-compiler/SKILL.md) — Batch prompt generation with shared visual DNA
- [Designly Director](../designly-director/SKILL.md) — Multi-asset campaign orchestration

### Schemas & References
- [Visual DNA Schema](schemas/visual-dna.schema.json) — Local DNA schema
- [Campaign Visual DNA](../../shared/references/campaign-visual-dna.md) — Continuity guide
- [Platform and Format Guide](../../shared/references/platform-and-format.md) — Safe zones & aspect ratios
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh handoff

Referenced files: 4

View saved version →

---
name: composition-director
description: Spatial composition, grid layout, hierarchy, and structural preflight director. This skill should be used when establishing layout grids, visual hierarchy, focal anchors, negative space balance, eye paths, or conducting preflight tests to prevent equal-emphasis slop and clutter.
---

# Composition Director

Composition Director establishes the mathematical and perceptual skeleton of the visual before generation or rendering. It governs grid systems, focal hierarchy, eye paths, scale relationships, and negative space balance.

---

## 1. Core Workflow

1. **Format & Canvas Setup**:
   - Establish aspect ratio (e.g. 1:1, 4:5, 9:16, 16:9) and define outer margins and safety padding (minimum 6-8% margin).
   - Select grid system: Rule of Thirds, Golden Ratio, Diagonal Dynamic Grid, or Swiss Modular Grid.

2. **Focal Anchor Architecture**:
   - **Primary Focal Anchor (Single)**: The hero element capturing immediate attention (occupies primary visual weight).
   - **Secondary Anchor**: Supporting narrative element or product context.
   - **Tertiary Elements**: Text blocks, badge, brand mark, and secondary accents.
   - **Strict Rule**: Exactly ONE primary focal point. Reject "equal-emphasis" where 3+ elements compete equally.

3. **Perceptual Preflight Checks**:
   - **The Thumbnail Test**: Scaled down to 100x100px, is the silhouette and hierarchy immediately clear?
   - **The Grayscale Contrast Test**: Stripped of color, do value contrasts maintain readable depth and clear separation?
   - **Negative Space Audit**: Ensure negative space is active and intentional, not trapped dead space.
   - **Accidental Tangencies**: Ensure overlapping elements have decisive overlap or clear separation; eliminate ambiguous touching edges.

4. **Output Contract**:
   - Return structured `composition_state` inside `DesignSignalPacket` containing grid specs, anchor coordinates, and negative space ratios.

---

## 2. Tools & Scripts

- Run design lint on Art Direction spec:
  ```bash
  python3 scripts/design_lint.py assets/art-direction.template.json
  ```
- Run design structural evals:
  ```bash
  python3 scripts/run_design_evals.py
  ```

---

## 3. Cross-Skill Neural Connections & References

### Peer & Specialist Skills
- [Typography Director](../typography-director/SKILL.md) — Text zone planning & typographic hierarchy
- [Arabic RTL Director](../arabic-rtl-director/SKILL.md) — Inverted RTL reading paths & Arabic layout balance
- [Photography Director](../photography-director/SKILL.md) — Camera focal length, perspective & lighting depth
- [Manipulation Director](../manipulation-director/SKILL.md) — Compositing scale, contact shadows & occlusion
- [Visual Storytelling](../visual-storytelling/SKILL.md) — Narrative storyboard arcs and eye path pacing
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and lock manager

### Schemas & References
- [Art Direction Schema](schemas/art-direction.schema.json) — Local spec schema
- [Creative Brief Schema](schemas/creative-brief.schema.json) — Local brief schema
- [Layout, Grid, and Spacing](../../shared/references/layout-grid-and-spacing.md) — Spatial rules
- [Composition and Photography](../../shared/references/composition-and-photography.md) — Perceptual mechanics
- [Design Preflight](../../shared/references/design-preflight.md) — Thumbnail and contrast checks
- [Gestalt and Perception](../../shared/references/gestalt-and-perception.md) — Visual grouping principles
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-bias composition

Referenced files: 9

View saved version →

---
name: creative-director
description: AI creative director with recursive self-assessment, Cannes-calibrated scoring, SIT/TRIZ structural ideation, and campaign canon anti-derivative preflight. This skill should be used when the user asks to generate creative concepts, brainstorm campaign ideas, develop a Big Idea or campaign platform, evaluate or critique existing creative work, find consumer insights, or shares a brief for ideation — including activations, PR-stunts, brand utility, experiential, and non-advertising ideas.
---

# Creative Director

Act as a creative director at the level of Droga5/Wieden+Kennedy/Mother. Core principle: insight before ideas. Use structural methodologies instead of free association. Be honest in evaluation, kill mediocrity, and apply Simplicity as Violence: the best ideas can be explained in one sentence.

Creativity = novelty + usefulness. Ultra-novel but useless = not creative. Generic and on-brief = also not creative. Find the intersection of the unexpected and the strategically precise.

## Instructions

### Phase Router

Determine the phase from context:

- New brief / request / "come up with" / "develop a concept" → start with **Phase 1: INTAKE**
- "Find an insight" / "what's behind this" / have a brief but no insight → **Phase 2: INSIGHT**
- "Generate ideas" / have an insight, need concepts → **Phase 3: IDEATION**
- "Evaluate the idea" / "improve the concept" / "critique" → **Phase 4: EVALUATE + REFINE**
- "Finalize" / "prepare a presentation" → **Phase 5: ARTICULATE**
- Full cycle (standard request) → sequentially Phase 1 → 2 → 3 → 4 → 5

---

### Phase 1: INTAKE (brief reception)

Extract from incoming material:
- Product/brand, category
- Target audience (who makes the decision? age, income, what frustrates them?)
- Business objective and communication objective
- Constraints (budget, channels, timelines, tone of voice, must-have elements)
- Competitive context
- Required idea level: Big Idea / Campaign Idea / Execution Idea

If data is insufficient, ask 3-5 precise questions. Not "tell me about the TA," but "who makes the purchase decision? age, income, main pain point?"

Determine the required idea level using the **Pollard 7-level taxonomy** (full reference: [Idea Taxonomy](references/idea-taxonomy.md)):

| Level | When required | Lifespan |
|---|---|---|
| `business` | new venture, repositioning the entire company | years |
| `brand` | rebranding, brand platform, "what does the brand stand for?" | 5-10+ years |
| `tagline` | short phrase that crystallizes brand idea | 5-10+ years |
| `advertising` | central thought across all comms — recognizable without logo | 3-5 years |
| `campaign` | seasonal campaign, product launch, promo | 3-12 months |
| `non_advertising` | activation/utility/cultural object that lives without ads | varies |
| `execution` | one-off channel/format/mechanic | days-weeks |

**Activation diagnostic:** if brief mentions activation/stunt/utility — apply the test "remove the campaign, does it still have meaning?" → Yes = `non_advertising` / No = `execution`. See [Activation Toolkit](references/activation-toolkit.md).

A `business` idea for shelf talkers = waste. An `execution` for rebranding = falling short. Mismatch is the #1 cause of creative-meeting friction.

---

### Phase 2: INSIGHT (insight discovery)

Load: [Insight Mining](references/insight-mining.md)

Sequence:

1. **Mark Pollard Four Points**: Problem → Insight → Advantage → Strategy
2. **JTBD**: what "job" does the consumer hire the communication for?
3. **Tension Spotting**: find one of three tensions:
   - Cultural (what society says vs what it does)
   - Category (what the category promises vs what it delivers)
   - Human (what a person wants vs what stands in the way)
4. **HMW**: 3 formulations at different levels of abstraction (broad / medium / narrow)
5. **Abstraction Laddering**: choose the optimal "rung" between abstract and concrete

**Insight quality test:** "Does this refresh one's view of the world? Does the person hear it and say 'yes, exactly, but I've never put it that way'?"

**Insight format:** one sentence: "[audience] wants [X], but [Y stands in the way], because [Z]"

---

### Phase 3: IDEATION (idea generation)

Load: [Methods Catalog](references/methods-catalog.md) + [Method Selection Matrix](references/method-selection-matrix.md)

For storytelling tasks additionally: [Storytelling Frameworks](references/storytelling-frameworks.md)

**Algorithm:**

0. **Prime against the canon.** Before generating, open the MOC most relevant to the brief context — [MOC Industry](references/legendary-campaigns/MOC-industry.md) (industry match), [MOC Budget](references/legendary-campaigns/MOC-budget.md) (budget constraint), or [MOC Emotion](references/legendary-campaigns/MOC-emotion.md) (emotional intent). Scan 5-7 canonical cases. Goal is anti-derivative: see what already exists in this slice so generation aims at the gap, not the pattern. Combining or remixing existing ideas across categories is allowed and encouraged.

1. Using [Method Selection Matrix](references/method-selection-matrix.md), select 3 methods from different categories:
   - One structural (SIT, SCAMPER, TRIZ, Morphological)
   - One association/collision (Bisociation, Random Entry, Synectics, Forced Connections)
   - One inversion/perturbation (Reverse Brainstorming, Worst Idea, Provocation PO, Oblique Strategies)

2. Generate 8-12 ideas, applying each method

3. Mark the first 3 ideas as **"conventional warmup"** (serial order effect: later ideas are statistically more original). Don't delete them, but bias toward ideas 5-12+

4. Each idea is tied to a specific insight/tension from Phase 2

5. Each idea is formulated in one sentence + 2-3 lines of development

6. **Tension test:** for each idea, check whether it carries an unresolved tension (cultural / category / human). If everything resolves cleanly → originality is weak. The best work lives in the unresolved gap. See [Legendary Patterns](references/legendary-patterns.md).

---

### Phase 4: EVALUATE + REFINE (recursive cycle)

Load: [Scoring Calibration](references/scoring-calibration.md) + [Creative Constitution](references/creative-constitution.md)

#### PASS 0: Idea Level Check

Before evaluation, verify: does the level of generated ideas match the `idea_type` requirement from Phase 1? Use the full Pollard 7-level taxonomy from [Idea Taxonomy](references/idea-taxonomy.md):

- `business` / `brand` — must scale for years, must answer "what does the company stand for?"
- `tagline` — must compress brand idea into ≤5 words
- `advertising` — central thought recognizable across channels for 3-5 years
- `campaign` — time-limited but expandable across channels
- `non_advertising` — must pass "remove the campaign, does it still mean something?" test
- `execution` — specific and implementable

#### PASS 1: Three-axis evaluation

**Axis 1: Brief Compliance (pass/fail)**
8 questions: Is there an idea? Conveys message? Responds to insight? Suits TA? Mandatory elements? Legal/ethical? Brand voice? Supported by attributes?

**Axis 2: Idea Strength (6 weighted criteria)**

| Criterion | Weight | What is evaluated |
|---|---|---|
| Originality | 0.25 | Unexpected? Empirical check in [MOC Pattern](references/legendary-campaigns/MOC-pattern.md). Saturated patterns capped at 6 unless structurally new. |
| Strategic fit | 0.20 | Solves brief objective and hits target audience. |
| Emotional response | 0.20 | Specificity via [Emotion Hierarchy](references/emotion-hierarchy.md) (Tier 1/2/3). Score 9+ requires Tier 3. |
| Feasibility | 0.15 | Implementable within budget/timeline/constraints. |
| Scalability | 0.10 | Series? Other media? Other markets? |
| Simplicity | 0.10 | Explainable in 10 seconds? One sentence? |

Weighted sum (1-10) = Score. In parallel: **HumanKind Score** (1-10).

**Axis 3: Scalability (4 questions)**
Longevity, abstraction rungs, cross-channel deployment, unified system.

#### PASS 2: Targeted improvement (if top < 9.0)
Identify weak criteria (<8), rotate to a different method from [Methods Catalog](references/methods-catalog.md), recalculate scores.

#### PASS 3-5: Deep improvement or restart
- Score >= 9.0 AND HumanKind >= 7 → run Pre-Mortem, exit to Phase 5
- Score 7.0-8.9 → continue with new method
- Score < 7.0 OR plateau → restart with case-soaking across [MOC Index](references/legendary-campaigns/MOC-index.md)

---

### Phase 5: ARTICULATE (final output)

Load: [Output Templates](assets/output-templates.md)

Final deliverable formatted as:
- Full cycle → **Top-3 Presentation Format**
- One idea in detail → **Creative Concept One-Pager**
- Strategic platform → **Campaign Platform**

---

## Output Contract & Neural Mesh Integration

When acting within the Designly Neural Mesh, Creative Director outputs a typed `DesignSignalPacket`:
- **strategy_state**: Insight formula, core tension, and chosen idea taxonomy level.
- **primary_message**: The single undeniable communication premise.
- **desired_action**: Cognitive or behavioral viewer reaction.
- **decisions**: Methodological lineage, canonical reference citations, and Cannes/HumanKind calibration scores.
- **recommended_next**: `["composition-director", "brand-intelligence", "taste-engine"]`

---

## Cross-Skill Neural Connections & References

### Peer & Specialized Skills
- [Insight Mining](../insight-mining/SKILL.md) — Upstream consumer tension spotting & JTBD
- [Campaign Canon](../campaign-canon/SKILL.md) — 571 canonical campaigns & P01-P18 saturation analysis
- [Brand Activation](../brand-activation/SKILL.md) — Experiential stunts & non-advertising diagnostics
- [Visual Storytelling](../visual-storytelling/SKILL.md) — Narrative frameworks & storyboard pacing
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and conflict resolver

### Domain Standards & Contracts
- [Methods Catalog](references/methods-catalog.md) — 20+ creative ideation methods (SIT, TRIZ, SCAMPER, Bisociation)
- [Method Selection Matrix](references/method-selection-matrix.md) — Method selection and rotation matrix
- [Scoring Calibration](references/scoring-calibration.md) — Cannes, D&AD, HumanKind, and Grey calibration rubrics
- [Creative Constitution](references/creative-constitution.md) — 3-layer critique constitution and diagnostic questions
- [Storytelling Frameworks](references/storytelling-frameworks.md) — 6 narrative frameworks (Story Spine, Sparkline, Pixar)
- [Insight Mining Guide](references/insight-mining.md) — Pollard Four Points, JTBD, Tension Spotting, Abstraction Laddering
- [Idea Taxonomy](references/idea-taxonomy.md) — Pollard 7-level taxonomy and activation diagnostic
- [Emotion Hierarchy](references/emotion-hierarchy.md) — Tier 1/2/3 emotional specificity hierarchy
- [Activation Toolkit](references/activation-toolkit.md) — 9 activation formats and execution diagnostics
- [Legendary Patterns](references/legendary-patterns.md) — P01-P18 pattern map and pre-mortem calibration
- [Legendary Campaigns Canon Index](references/legendary-campaigns/MOC-index.md) — 571 verified canonical campaigns
- [Output Templates](assets/output-templates.md) — Presentation templates and one-pagers
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-bias standards
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh signal contract
- [Design Context Schema](../../shared/contracts/design-context.schema.json) — Shared state context

---

*Attribution: Serge Shima ([aimasters.me](https://aimasters.me)) · Licensed under CC BY 4.0.*

Referenced files: 597

View saved version →

---
name: creative-strategy
description: Marketing strategy and visual concept specialist. This skill should be used when deconstructing marketing briefs, defining target audience insights, establishing message hierarchy, developing distinct concept territories, or clarifying the single primary communication job before visual design begins.
---

# Creative Strategy

Creative Strategy deconstructs commercial marketing briefs into sharp communication objectives, audience insights, message hierarchies, and distinct visual concept territories.

---

## 1. Core Workflow

1. **Brief Deconstruction**:
   - Extract the core business objective (e.g. product launch, brand repositioning, direct conversion).
   - Identify the primary target persona (demographic, psychographic, pain points, motivations).
   - Isolate the single **Primary Communication Job** that must be communicated within 3 seconds.

2. **Message Hierarchy**:
   - **Primary Message**: The single indispensable claim or emotion (occupies 100% of initial viewer glance).
   - **Secondary Message**: Supporting reason-to-believe or value proposition.
   - **Tertiary Elements**: Call to action (CTA), legal disclaimers, or secondary brand proof points.

3. **Concept Territory Development**:
   - For exploratory briefs, define up to 3 genuinely distinct visual territories (e.g. Territory A: Minimalist Precision, Territory B: Organic Authenticity, Territory C: Dynamic Motion).
   - Never generate variations that are merely superficial color swaps.

4. **Output Contract**:
   - Return a valid `DesignSignalPacket` containing `strategy_state`, `primary_message`, and `desired_action` decisions with supporting evidence.

---

## 2. Rules & Hard Constraints

- **Single Core Job**: If a brief demands 5 simultaneous headlines, force a hierarchy: 1 primary, 2 secondary, 2 relegated to caption/metadata.
- **Audience Grounding**: Every visual metaphor must resonate with the target demographic's cultural and commercial context.
- **No Premature Execution**: Strategy does not dictate camera lens millimeter choices or hexadecimal color codes; it defines conceptual intent and hierarchy.

---

## 3. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Insight Mining](../insight-mining/SKILL.md) — Unearthing underlying consumer tensions & JTBD
- [Creative Director](../creative-director/SKILL.md) — Structural ideation (SIT/TRIZ) & Cannes scoring
- [Campaign Canon](../campaign-canon/SKILL.md) — Benchmarking against 571 canonical campaigns
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Aligning strategic territories with brand identity
- [Composition Director](../composition-director/SKILL.md) — Translating message hierarchy into layout structure
- [Designly Director](../designly-director/SKILL.md) — Orchestrator and lock manager

### Schemas & References
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh handoff
- [Marketing Brief Guide](../../shared/references/marketing-brief.md) — Brief intake rules
- [Design Principles](../../shared/references/design-principles.md) — Core principles
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-stereotype audience modeling

Referenced files: 3

View saved version →

---
name: designly-director
description: Lead commercial Art Director and Design Neural Mesh orchestrator. This skill should be used when orchestrating end-to-end commercial design, resolving conflicting brand/taste/structure signals, locking brief constraints, delegating specialist analysis, sanitizing bounded edits, and conducting final art-direction signoff.
---

# Designly Director

Designly Director is the primary orchestrator for Designly. It coordinates 19 specialist Skills and 14 custom Codex agents through typed handoffs, enforces immutable locks, and routes failing dimensions during revision.

```text
USER BRIEF / FEEDBACK
    |
    v
[designly-director]
    |
    +--> [creative-director] (Cannes/HumanKind calibration, SIT/TRIZ)
    +--> [insight-mining] (Tension spotting & Pollard 4-points)
    +--> [campaign-canon] (571 legendary campaigns & pattern benchmarking)
    +--> [brand-activation] (PR stunts, brand utility, experiential)
    +--> [visual-storytelling] (Narrative arcs, Story Spine, Sparkline)
    +--> [creative-strategy] (Brief deconstruction, message hierarchy)
    +--> [brand-intelligence] (Brand rules & logo audit)
    +--> [taste-engine] <--> [reference-memory] (Transferable rules & REF IDs)
    |
    v
[composition-director]
    |
    +--> [typography-director] --> [arabic-rtl-director] when needed
    +--> [photography-director]
    +--> [manipulation-director]
    +--> [campaign-dna]
    |
    +--> if EXISTING IMAGE + bounded edit/annotation/inpainting
    |        [edit-sanitizer] --> ready EditContract
    |
    v
[prompt-compiler]
    |
    v
IMAGE EXECUTION
    |
    v
[visual-qa]
    +--> PASS: signoff
    +--> FAIL: one RevisionRequest to responsible specialist
```

## Responsibilities

1. Parse the job, deliverables, assets, platform and explicit constraints
2. Turn user and documented brand constraints into high-priority locks
3. Delegate independent read-heavy analysis when the host supports subagents
4. Fall back to the same Skills sequentially when subagents are unavailable
5. Merge signals by authority rather than by confidence alone
6. Refuse specialist usurpation: use the dedicated Skill for ideation, composition, type, craft, editing safety or QA
7. On edits, preserve source lineage and route raw correction notes through `edit-sanitizer` before `prompt-compiler`
8. On QA failure, rerun only the responsible node unless the concept itself failed

## Signal priority

1. user exact constraints
2. documented brand/product rules
3. safety, cultural and exact-copy hard gates
4. primary communication job
5. hierarchy and composition
6. accessibility and legibility
7. campaign continuity
8. craft realism
9. explicit user taste
10. inferred taste
11. decorative finish

A lower-priority signal never overwrites a higher-priority lock

## Intake

Create or update `DesignContext`

- task type
- objective
- audience
- primary message
- desired action
- platform and aspect ratio
- cultural/language context
- supplied assets
- locks

Ask a question only when the unresolved detail materially changes the result or a sanitizer returns `clarify`

## Generation path

1. Creative direction (Cannes calibration, tension mining, SIT/TRIZ ideation, canon benchmarking), strategy and brand/taste analysis
2. Structural composition
3. Typography/Arabic and craft specialists as applicable
4. Campaign DNA for multi-asset work
5. Prompt Compiler
6. Image execution
7. Visual QA

## Existing-image edit path

Do not treat correction notes as a normal generation prompt

1. Identify the last approved source checkpoint
2. Collect annotation geometry or semantic target plus source dimensions
3. Route the raw edit request to `edit-sanitizer`
4. If `clarify`, ask exactly one target/scope question
5. If `reject` or `veto`, do not call an image editor
6. If `ready`, store the EditContract in `edit_state`
7. If Arabic copy is changing, run `arabic-rtl-director` before execution
8. Send only the ready EditContract to `prompt-compiler`
9. Execute against the approved source checkpoint
10. Visual QA inspects target accuracy and collateral drift
11. If the edit fails, retry from the approved source, never from the failed render
12. Stop after three failed attempts and report the persistent defect rather than accumulating drift

## Revision routing

- concept/originality/ideation -> `creative-director`
- insight depth/tensions -> `insight-mining`
- pattern saturation/canon -> `campaign-canon`
- activation mechanics -> `brand-activation`
- narrative arc/storyboard -> `visual-storytelling`
- strategy/audience/brief -> `creative-strategy`
- hierarchy/composition -> `composition-director`
- typography -> `typography-director`
- Arabic/glyphs -> `arabic-rtl-director`
- brand/product -> `brand-intelligence`
- physical integration -> `manipulation-director`
- annotation mapping/edit scope/collateral drift -> `edit-sanitizer`
- provider instruction mismatch -> `prompt-compiler`

## Final signoff

Never approve from an Art Direction Spec alone when an actual visual exists. Inspect the actual output. Approval requires weighted QA threshold, applicable category floors, hard gates, slop veto, and for edits a passed edit-scope/collateral-change review.

---

## Cross-Skill Neural Connections & References

### Strategic & Ideation Upstream
- [Creative Director](../creative-director/SKILL.md) — Cannes/D&AD calibration and recursive ideation
- [Insight Mining](../insight-mining/SKILL.md) — Consumer tensions, JTBD, and Pollard 4-points
- [Campaign Canon](../campaign-canon/SKILL.md) — 571 canonical campaigns & P01-P18 pattern taxonomy
- [Brand Activation](../brand-activation/SKILL.md) — Experiential stunts & non-advertising diagnostics
- [Visual Storytelling](../visual-storytelling/SKILL.md) — Narrative frameworks & emotional tiers
- [Creative Strategy](../creative-strategy/SKILL.md) — Audience personas and communication objectives
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Brand guidelines and product fidelity
- [Taste Engine](../taste-engine/SKILL.md) — Reference deconstruction and transferable rules

### Craft & Downstream Execution
- [Composition Director](../composition-director/SKILL.md) — Layout, grid, hierarchy, negative space
- [Typography Director](../typography-director/SKILL.md) — Type hierarchy, measure, line breaks
- [Arabic RTL Director](../arabic-rtl-director/SKILL.md) — RTL visual flow and calligraphy glyph checks
- [Photography Director](../photography-director/SKILL.md) — Camera optics, lighting, and materials
- [Manipulation Director](../manipulation-director/SKILL.md) — Compositing physics and realism
- [Campaign DNA](../campaign-dna/SKILL.md) — Multi-asset visual family continuity
- [Edit Sanitizer](../edit-sanitizer/SKILL.md) — Bounded edit and inpainting protection
- [Prompt Compiler](../prompt-compiler/SKILL.md) — Provider-ready generation and edit instructions
- [Visual QA](../visual-qa/SKILL.md) — Independent visual review, floors, and revision routing

### Contracts & Standards
- [Software Architecture](../../shared/references/software-architecture-and-contracts.md) — Bounded contexts & lock precedence
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-bias & authentic human representation
- [Design Context Schema](../../shared/contracts/design-context.schema.json) — Neural Mesh shared state
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Specialist signal handoff
- [Routing Graph](../../shared/contracts/routing-graph.json) — Complete mesh topology

Referenced files: 3

View saved version →

---
name: edit-sanitizer
description: Pre-execution sanitizer for annotation-guided edits, inpainting, copy corrections, local retouching, and object replacement. This skill should be used when a user points at or annotates part of an existing image and expects a bounded change without collateral redesign, or when edit scope, mask geometry, exact copy, or protected regions must be validated before image execution.
---

# Edit Sanitizer

Edit Sanitizer is the fail-close boundary between user feedback/annotations and image-edit execution. It exists because generative image editors may reinterpret more of a source image than the user intended. Do not send a local-edit request directly to `prompt-compiler` until this Skill has produced a `ready` EditContract.

## Core rule

Convert feedback into the smallest explicit mutation that can satisfy the request.

Do not treat a scribble, arrow, selection, bounding box, or phrase such as `fix this` as self-explanatory when more than one plausible target exists.

## Workflow

1. Identify the approved source checkpoint
2. Resolve annotation coordinates against the source dimensions
3. Map the annotation to one semantic target
4. Split user feedback into atomic requested mutations
5. Reject or veto unrelated global restyling when the request says `only`, `just`, `this area`, or otherwise implies a bounded edit
6. Lock crop, canvas, perspective, identities, exact text, lighting, and unaffected composition unless the user explicitly targets one of them
7. Derive protected regions from the complement of the editable target plus semantic locks
8. Set a conservative mutation budget
9. Require Arabic review for Arabic copy corrections
10. Produce a typed EditContract
11. Only when status is `ready`, forward the contract to `prompt-compiler`

## Annotation handling

Treat annotation geometry and semantic intent as separate evidence:
- `bbox`: validate positive area and source bounds
- `polygon`: require a valid polygon around the intended area
- `mask_ref`: require an actual mask reference
- `semantic`: require one clearly identifiable object or text region
- `normalized`: convert coordinates to native pixels before execution

If two plausible targets have similar confidence, return `clarify` rather than guessing.

## Drift prevention

Every retry starts from `source_checkpoint`, which must be the last approved source.

Never chain a second corrective edit from a failed or visibly drifted render.

Maximum edit loop: 3 attempts. If the same defect persists or collateral changes grow, stop and return the evidence rather than continuing destructive edits.

## Bounded-edit vetoes

Veto execution when:
- a local edit also asks to restyle the whole image
- there are multiple unrelated mutations that should be separate edits
- the target is outside the source or has zero area
- the annotation cannot be mapped confidently
- exact replacement copy is missing for copy correction
- the retry source is not the approved checkpoint

## Preservation language

Do not promise literal pixel identity from a generative editor. Require material stability outside the edit target and validate collateral changes after execution.

## Output

Return the EditContract plus a short execution verdict:
- `ready`: may proceed to Prompt Compiler
- `clarify`: ask one target/scope question, do not execute
- `reject`: invalid geometry/source/copy contract, do not execute
- `veto`: conflicting or over-broad mutation request, do not execute

## Runtime helper

```bash
python3 ../../shared/scripts/sanitize_edit.py < edit-request.json
```

---

## Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Prompt Compiler](../prompt-compiler/SKILL.md) — Downstream consumer of sanitized EditContracts
- [Manipulation Director](../manipulation-director/SKILL.md) — Boundary blending physics & light wrap
- [Arabic RTL Director](../arabic-rtl-director/SKILL.md) — Text ligature verification on Arabic copy edits
- [Visual QA](../visual-qa/SKILL.md) — Target accuracy & collateral drift verification
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and edit state manager

### Schemas & References
- [Edit Contract Schema](../../shared/contracts/edit-contract.schema.json) — Bounded edit schema
- [Image Editing Guidance](../../shared/references/image-editing.md) — Editing rules
- [Routing Graph](../../shared/contracts/routing-graph.json) — Mesh topology & feedback loops

Referenced files: 3

View saved version →

---
name: insight-mining
description: Specialist in unearthing consumer tensions and strategic insight formulation. This skill should be used when discovering deep audience insights, mapping Jobs-To-Be-Done (JTBD), identifying cultural, category, or human tensions, applying Mark Pollard's Four Points, or laddering abstraction levels before creative ideation begins.
---

# Insight Mining

Insight Mining transforms ambiguous audience data and business problems into singular, undeniable human insights. Great advertising does not begin with an execution; it begins with an insight that refreshes the viewer's understanding of their own experience.

---

## 1. Core Workflow

1. **Mark Pollard Four Points**:
   - **Problem**: What is genuinely broken or difficult for the audience?
   - **Insight**: The unstated human or cultural truth behind the problem.
   - **Advantage**: The brand's unique point of difference that resolves or illuminates the tension.
   - **Strategy**: The overarching bridge connecting the advantage to the consumer's emotional reality.

2. **Tension Spotting**:
   Identify at least one of three load-bearing tensions:
   - **Cultural Tension**: What society proclaims vs what people actually experience.
   - **Category Tension**: What the industry promises vs what it realistically delivers.
   - **Human Tension**: What an individual desires vs what emotional or physical barrier stands in the way.

3. **Jobs-To-Be-Done (JTBD)**:
   - What functional, emotional, or social job is the consumer "hiring" this communication to perform?

4. **Abstraction Laddering & HMW Formulations**:
   - Create 3 "How Might We" formulations across abstraction levels:
     - Broad (macro cultural challenge)
     - Medium (category behavioral shift)
     - Narrow (immediate tactical reaction)

5. **Insight Synthesis**:
   - Formulate the insight in one crisp sentence:
     `"[Audience] wants [desire X], but [barrier Y stands in the way], because [root truth Z]."`

---

## 2. Quality & Validation Gates

- **The Refresh Test**: When the target audience hears this insight, do they say: *"Yes, exactly, but I've never put it into words before"*?
- **Not a Demographic Stat**: Demographics are facts, not insights. An insight is an unspoken emotional truth.
- **Anti-Banal Filter**: If every competitor in the category already uses this observation, it is an observation, not an insight.

---

## 3. Cross-Skill Neural Connections & References

### Peer & Specialized Skills
- [Creative Director](../creative-director/SKILL.md) — Downstream SIT/TRIZ ideation and Cannes scoring
- [Creative Strategy](../creative-strategy/SKILL.md) — Target persona and brief deconstruction
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Core brand values and identity bounds
- [Visual Storytelling](../visual-storytelling/SKILL.md) — Narrative emotional tiers
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator

### Domain References & Contracts
- [Insight Mining Guide](references/insight-mining.md) — Comprehensive Mark Pollard & JTBD methodology
- [Idea Taxonomy](references/idea-taxonomy.md) — 7-level taxonomy and strategic alignment
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-bias standards
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh signal handoff
- [Design Context Schema](../../shared/contracts/design-context.schema.json) — Shared state context

Referenced files: 6

View saved version →

---
name: manipulation-director
description: Digital manipulation, compositing physics, perspective alignment, and photo-integration director. This skill should be used when combining multiple image elements, planning compositing perspective and scale, enforcing realistic contact shadows and reflections, or directing impossible/surreal advertising scenes with internal physical consistency.
---

# Manipulation Director

Manipulation Director governs the physical plausibility and visual integration of composite scenes, multi-element photo-manipulations, and surreal commercial concepts. It ensures that impossible ideas obey rigorous internal physics so they look tactile and believable rather than pasted-together AI slop.

---

## 1. Core Workflow

1. **Perspective & Vanishing Point Alignment**:
   - Establish the primary horizon line and camera eye level across all integrated elements.
   - Ensure converging perspective lines of inserted objects match the background plate precisely.

2. **Contact Shadows & Grounding Physics**:
   - **Ambient Occlusion Shadow**: Ultra-dark, sharp, thin contact line where the object touches the ground surface.
   - **Cast Shadow**: Directional shadow extending away from the primary light source with realistic penumbra (sharp near base, softening with distance).
   - **Secondary Bounce**: Color bleeding from the ground onto the underside of the object.

3. **Directional Reflections & Fresnels**:
   - On glossy, water, or metallic surfaces, calculate accurate reflection angles based on camera perspective.
   - Apply Fresnel falloff (reflections strengthen at grazing angles).

4. **Environment Relighting & Light Wrap**:
   - Relight inserted elements to match the color temperature, intensity, and key direction of the host environment.
   - Apply subtle atmospheric light wrap at the high-contrast silhouetted boundaries of the subject.

5. **Internal Consistency for Surreal Concepts**:
   - If the concept is physically impossible (e.g., a massive floating perfume bottle over desert dunes), treat the impossible element as a real physical entity casting authentic shadows, reflecting the sky, and interacting plausibly with environment dust and light.

6. **Output Contract**:
   - Return structured `craft_state.manipulation` inside `DesignSignalPacket`.

---

## 2. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Photography Director](../photography-director/SKILL.md) — Lighting angles, color temperature, and camera focal lengths
- [Edit Sanitizer](../edit-sanitizer/SKILL.md) — Inpainting boundaries, polygon geometry, and protected areas
- [Composition Director](../composition-director/SKILL.md) — Spatial balance, scale relationships, and grid anchors
- [Brand Activation](../brand-activation/SKILL.md) — Public ambient stunts and physical mockups
- [Visual QA](../visual-qa/SKILL.md) — Physical believability floor & artifact inspection
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and lock manager

### Schemas & References
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Signal handoff contract
- [Advertising Manipulation](../../shared/references/advertising-manipulation.md) — Compositing physics & realism
- [Image Editing Guide](../../shared/references/image-editing.md) — Editing methods & source protection

Referenced files: 3

View saved version →

---
name: photography-director
description: Commercial photography, studio lighting, camera optics, and material physics director. This skill should be used when directing camera focal length, depth of field, 3-point studio lighting systems, shutter speed, color temperature, or realistic surface material finishes.
---

# Photography Director

Photography Director specifies realistic photographic optics, camera physics, studio lighting environments, and physical material behaviors. It applies photography language purposefully when real optical qualities are required.

---

## 1. Core Workflow

1. **Camera Optics & Lens Choice**:
   - **Wide (24mm - 35mm)**: Environmental context, architectural space, dramatic perspective.
   - **Standard (50mm)**: Natural human eye perspective, documentary realism, undistorted geometry.
   - **Telephoto / Portrait (85mm - 135mm)**: Compressed perspective, flattering facial proportions, clean product packshots, shallow depth-of-field separation.
   - **Macro (100mm+)**: Extreme texture detail, jewelry, watch dial craftsmanship, microscopic surfaces.

2. **Aperture & Depth of Field**:
   - `f/1.4 - f/2.8`: Pronounced optical bokeh, subject isolation, soft dreamy falloff.
   - `f/5.6 - f/8.0`: Commercial product standard, tack-sharp edge-to-edge subject clarity with natural background separation.
   - `f/11 - f/16`: Deep depth-of-field for landscapes and complex architectural scenes.

3. **3-Point Studio Lighting Architecture**:
   - **Key Light**: Primary illumination source, defines form, mood, and directional shadow angle (e.g. 45° overhead softbox).
   - **Fill Light**: Controls shadow contrast ratio (e.g. 2:1 for high-key commercial, 8:1 for moody dramatic).
   - **Rim / Kicker Light**: Edge illumination separating dark subjects from dark backgrounds.
   - **Color Temperature**: Specify Kelvin values (e.g. 3200K warm tungsten, 5600K clean daylight, 6500K cool overcast).

4. **Surface Material Physics & Inclusive Lighting**:
   - Specify material response accurately: diffuse reflection, specular roughness, anisotropy (brushed aluminum), translucency (subsurface scattering on skin/liquids), and index of refraction (glass/water).
   - Melanin-accurate color grading: avoid ashen undertones or blown-out highlights on deeper complexions.

5. **Output Contract**:
   - Return structured `craft_state.photography` inside `DesignSignalPacket`.

---

## 2. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Composition Director](../composition-director/SKILL.md) — Spatial framing, focal anchors, and scale relationships
- [Manipulation Director](../manipulation-director/SKILL.md) — Environmental lighting alignment in composites
- [Taste Engine](../taste-engine/SKILL.md) — Reference lighting extraction and aesthetic rules
- [Prompt Compiler](../prompt-compiler/SKILL.md) — Translation into provider optical syntax
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and lock manager

### Schemas & References
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Signal handoff contract
- [Composition and Photography](../../shared/references/composition-and-photography.md) — Photographic mechanics
- [Color and Contrast](../../shared/references/color-and-contrast.md) — Value contrast & color temperature
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Skin tone lighting & anti-bias

Referenced files: 3

View saved version →

---
name: prompt-compiler
description: Image-generation and image-edit instruction compiler. This skill should be used when translating an approved Art Direction Spec or a ready EditContract into precise provider/model-ready instructions, linting prompt slop, or preparing execution for the host image tool without changing upstream creative decisions.
---

# Prompt Compiler

Compile approved design decisions into executable image instructions. Do not invent the concept, repair weak hierarchy, reinterpret annotations, or broaden edit scope.

## Preconditions

### Generation

Require approved strategy/composition/craft state appropriate to the task

### Editing, inpainting, annotation-guided correction

Require a typed `EditContract` from `edit-sanitizer` with

- `status: ready`
- `execution_allowed: true`
- approved `source_checkpoint`
- normalized target geometry
- atomic requested mutations
- protected regions and semantic locks
- acceptance checks

If these are missing, route to `edit-sanitizer`. Do not compile a best-effort local edit from raw feedback

## Compilation rules

1. State subject identity and required invariants first
2. Express composition as relationships, not disconnected style words
3. Use camera language only when it materially changes the visual result
4. Preserve brand, product, copy, and campaign locks verbatim
5. Include only likely failure exclusions for the current task
6. Keep provider-specific syntax modular and subordinate to the Art Direction Spec
7. When the host has a native image tool, use its supported interface instead of inventing API parameters
8. Apply inclusive visual representation rules: eliminate clone faces, mandate authentic ethnic features and melanin-calibrated lighting

## Bounded edit compilation

Compile the sanitized contract in this order

1. Approved source checkpoint
2. Exact semantic target and normalized/native geometry if the host accepts region data
3. One atomic allowed mutation
4. Identity, geometry, text, and style locks
5. Protected-region rule for all non-target content
6. Boundary blending allowance limited to what is needed for believable integration
7. Acceptance checks
8. Retry rule: restart from the approved checkpoint if collateral drift is material

Use language such as

`Edit only the selected bottle-cap region. Change the cap finish from matte black to brushed silver. Keep the bottle silhouette, label, logo, crop, perspective, background, lighting direction, and color grade materially stable. Do not add, remove, move, restyle, or rewrite anything outside the target. Allow only the minimal edge blending needed to integrate the new cap finish. If the result changes protected content, reject it and retry from the approved source checkpoint.`

Do not claim literal pixel identity unless the available editor guarantees it deterministically

## Copy correction

- exact copy is immutable
- never ask the image model to invent replacement copy
- Arabic copy requires `arabic-rtl-director` review before execution
- if text correctness remains unreliable, generate/retain the visual foundation and use a deterministic text workflow where the host supports one

## Prompt lint

Run

```bash
python3 scripts/prompt_lint.py "<compiled instruction>"
python3 scripts/test_prompt_lint.py
```

Reject prompt text when vague quality adjectives or effect stacks replace visual decisions

## Output

Return `generation_state` with

- operation: `generate` or `edit`
- provider/host assumptions actually verified
- compiled instruction
- source checkpoint when editing
- edit contract ID when editing
- acceptance checks
- likely failure exclusions

---

## Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Edit Sanitizer](../edit-sanitizer/SKILL.md) — Upstream sanitized EditContract provider
- [Photography Director](../photography-director/SKILL.md) — Optical parameters and lighting models
- [Manipulation Director](../manipulation-director/SKILL.md) — Compositing physics and boundary blending rules
- [Visual QA](../visual-qa/SKILL.md) — Downstream independent visual verification gate
- [Designly Director](../designly-director/SKILL.md) — Orchestrator and state owner

### Schemas & References
- [Prompt Compiler Guide](../../shared/references/prompt-compiler.md) — Model prompt translation rules
- [Model Guides](../../shared/references/model-guides.md) — Provider-specific nuances
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-bias prompting rules
- [Edit Contract](../../shared/contracts/edit-contract.schema.json) — Local edit schema
- [Signal Packet](../../shared/contracts/signal-packet.schema.json) — Neural Mesh handoff

Referenced files: 5

View saved version →

---
name: reference-memory
description: Local-first reference memory and scoped preference ledger manager. This skill should be used when saving, recalling, updating, or deleting reference records with stable REF IDs (e.g. REF-1042), managing user likes/dislikes, or querying persistent taste preferences.
---

# Reference Memory

Reference Memory provides local-first, transparent persistence for user references, assigned design jobs, and scoped feedback ledgers. It uses stable human-readable IDs (`REF-####`) and does not make misleading claims of fine-tuning or modifying underlying model weights.

---

## 1. Core Workflow

1. **Reference Registration**:
   - Assign next available sequential ID: `REF-####` (e.g. `REF-1001`, `REF-1002`).
   - Store reference metadata, file path, source tags, and primary design jobs (e.g. `lighting`, `composition`, `palette`).

2. **Scoped Preference Feedback**:
   - When a user expresses a preference (e.g., "I love the rim light in REF-1042 but hate the typography"), scope the feedback precisely:
     - Record `likes: ["lighting.rim_light"]`
     - Record `dislikes: ["typography.layout"]`
   - Never allow a dislike on one dimension (typography) to discard or invalidate rules on another dimension (lighting).

3. **Recall & Filtering**:
   - Query references by job tag, domain (e.g. `fmcg`, `luxury`, `automotive`), or aesthetic keyword.
   - Return active Taste Profiles associated with the recalled reference.

4. **Output Contract**:
   - Emits structured JSON reference records and `DesignSignalPacket` containing `taste_state.reference_memory`.

---

## 2. CLI Tooling

Execute the local Reference Memory CLI:
```bash
# Add a new reference
python3 scripts/reference_memory.py add --id REF-1001 --title "Nordic Minimalist Poster" --jobs composition,palette

# Query existing references
python3 scripts/reference_memory.py list

# Recall specific reference
python3 scripts/reference_memory.py get REF-1001

# Record scoped feedback
python3 scripts/reference_memory.py feedback REF-1001 --like "lighting.soft_directional" --dislike "background.clutter"
```

---

## 3. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Taste Engine](../taste-engine/SKILL.md) — Extracting transferable rules from recalled references
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Auditing reference tags against brand constraints
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and memory state manager

### Schemas & References
- [Reference Memory Schema](schemas/reference-memory.schema.json) — Local schema
- [Reference Memory Guide](../../shared/references/reference-memory.md) — Usage rules
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh handoff

Referenced files: 7

View saved version →

---
name: taste-engine
description: Taste extraction and transferable visual rule mixing engine. This skill should be used when deconstructing visual reference images into transferable design principles, building Taste Profiles, mixing multiple references by assigned design jobs, or preventing plagiarism and AI-slop anti-patterns.
---

# Taste Engine

Taste Engine converts raw reference images into evidence-backed, transferable design rules and structured Taste Profiles. It prevents literal copying by abstracting principles (lighting ratios, color harmonies, focal geometry, surface finishes) and assigning dedicated jobs to each reference.

---

## 1. Core Workflow

1. **Reference Deconstruction**:
   - Analyze reference image across 5 core dimensions:
     - **Lighting**: Key angle, fill ratio, shadow softness, color temperature.
     - **Palette**: Dominant, secondary, and accent ratios (e.g. 60-30-10 rule).
     - **Texture & Finish**: Material response (matte, satin, metallic, specular).
     - **Spatial Geometry**: Grid anchors, visual weight distribution, negative space.
     - **Mood & Tension**: Atmosphere, grain, contrast level.

2. **The 4-Step Rule Extraction Pipeline**:
   - `Evidence` (what is physically visible in the image)
   - `Observation` (structural or aesthetic pattern)
   - `Transferable Rule` (generalizable principle applicable to new briefs)
   - `Constraint / Anti-Rule` (what to strictly avoid to prevent unintended slop)

3. **Job-Based Reference Mixing**:
   - Never blend references indiscriminately.
   - Assign distinct, non-overlapping design jobs:
     - Reference A -> **Lighting Job** (e.g. 45° soft directional rim lighting).
     - Reference B -> **Composition Job** (e.g. dynamic diagonal off-center anchor).
     - Reference C -> **Color Harmony Job** (e.g. muted warm earth tones).

4. **Originality & Anti-Plagiarism Guard**:
   - Strictly discard subject matter, character identity, and proprietary artistic trademarks from the reference.
   - Never copy unique composite artwork verbatim.

5. **Output Contract**:
   - Emits structured `TasteProfile` JSON and `DesignSignalPacket` containing `taste_state`.

---

## 2. Tools & Scripts

- Lint Taste Profile: `python3 scripts/taste_lint.py <taste-profile.json>`
- Merge References: `python3 scripts/taste_merge.py <taste-mix.json>`

---

## 3. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Reference Memory](../reference-memory/SKILL.md) — Persistent REF-ID lookup and taste history
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Checking reference rules against brand locks
- [Composition Director](../composition-director/SKILL.md) — Applying reference geometric anchors to grid
- [Photography Director](../photography-director/SKILL.md) — Translating lighting extractions into optics
- [Campaign Canon](../campaign-canon/SKILL.md) — Cross-referencing visual style with canon cases
- [Designly Director](../designly-director/SKILL.md) — Orchestrator and conflict resolver

### Schemas & References
- [Taste Profile Schema](schemas/taste-profile.schema.json) — Local profile schema
- [Taste Mix Schema](schemas/taste-mix.schema.json) — Mixing contract schema
- [Taste Engine Guide](../../shared/references/taste-engine.md) — Extraction principles
- [Reference Analysis](../../shared/references/reference-analysis.md) — Analytical steps

Referenced files: 11

View saved version →

---
name: typography-director
description: Typographic hierarchy, layout, measure, and exact-copy director. This skill should be used when setting typographic scale, headline measure, line breaks, text zones, contrast ratios, and layout boundaries, or protecting exact client copy strings.
---

# Typography Director

Typography Director governs typographic hierarchy, measure, line length, kerning, leading, and text legibility. It establishes dedicated, uncluttered text zones and protects exact user copy from being rewritten, fragmented, or lost against busy backgrounds.

---

## 1. Core Workflow

1. **Hierarchy & Scale Definition**:
   - **Headline**: High visual contrast, tight leading (1.05 - 1.15x), short line length (1-4 words per line).
   - **Subhead**: Supporting scale (0.4 - 0.6x headline size), comfortable leading (1.2 - 1.3x).
   - **Body / Callout**: High legibility, measure of 45-75 characters per line, generous leading (1.4 - 1.6x).
   - **CTA / Microcopy**: Distinct geometry, uppercase or balanced weight, isolated from decorative imagery.

2. **Semantic Line Breaking**:
   - Break lines strictly at natural grammatical and semantic boundaries.
   - Never leave single-word orphans or break hyphenated concepts awkwardly across lines.

3. **Text Zone & Contrast Protection**:
   - Designate clean, low-frequency background zones for text placement.
   - Ensure minimum contrast ratio:
     - Large text (headlines >= 24px): >= 3.0:1 (target >= 4.5:1).
     - Standard body text: >= 4.5:1 (target >= 7.0:1).
   - If the background is complex photographic imagery, specify solid scrims, soft directional vignettes, or clean flat containers behind the text.

4. **Arabic Copy Routing**:
   - If the text contains Arabic script, immediately route layout constraints to [Arabic RTL Director](../arabic-rtl-director/SKILL.md) to ensure proper right-to-left baseline flow and calligraphy rules.

5. **Output Contract**:
   - Return structured `typography_state` inside `DesignSignalPacket` containing font families, scale ratios, text zone coordinates, and exact copy locks.

---

## 2. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Arabic RTL Director](../arabic-rtl-director/SKILL.md) — RTL typography, glyph shaping & calligraphy fidelity
- [Composition Director](../composition-director/SKILL.md) — Grid placement, margins, and spatial hierarchy
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Corporate typeface rules and brand lock alignment
- [Prompt Compiler](../prompt-compiler/SKILL.md) — Text zone protection in image generation instructions
- [Designly Director](../designly-director/SKILL.md) — Orchestrator and exact-copy lock manager

### Schemas & References
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh handoff
- [Typography Guide](../../shared/references/typography.md) — Typographic hierarchy and measure
- [Design Principles](../../shared/references/design-principles.md) — Core visual design rules

Referenced files: 3

View saved version →

---
name: visual-qa
description: Independent visual quality assurance, hard-gate auditor, and targeted revision router. This skill should be used when reviewing generated or edited visuals, testing hierarchy/craft/brand fidelity, detecting AI slop, checking bounded-edit collateral drift, issuing approval verdicts, or routing targeted RevisionRequests.
---

# Visual QA

Review the actual visual independently. Do not defend upstream decisions and do not approve an uninspected hypothetical output.

## Core review

Evaluate applicable categories including:

- brief accuracy and primary message
- concept strength
- hierarchy and composition
- grouping, spacing and crop
- typography and exact copy
- color/contrast
- brand fidelity
- product fidelity
- physical believability
- cultural/platform fit and inclusive representation
- overall craft
- AI slop

Weighted score must meet the configured threshold and every applicable category floor must pass

## Hard gates

Fail regardless of average when applicable:

- required copy is wrong
- Arabic glyphs/connections or RTL behavior are malformed
- official logo/mark is malformed
- supplied product identity/proportions materially drift
- focal anatomy, human dignity, or physical interaction has a critical defect
- protected edit content changes materially outside what boundary blending requires
- local edit changes crop, canvas, camera, layout, lighting, text, identity or style that the EditContract locked
- annotation was applied to the wrong semantic target
- AI-slop veto triggers

## Bounded-edit review

When `edit_state` exists, compare output to the approved `source_checkpoint`, not to a previous failed edit. Evaluate target accuracy, edit-scope accuracy, and collateral change.

## Perception checks

- one-second hierarchy test
- thumbnail test
- squint/blur value-mass test
- grayscale hierarchy test
- edge/tangency/crop test
- brand-off specificity test
- 7-point inclusive representation review
- effect-subtraction test
- physics pass
- character-by-character copy pass
- source-vs-output edit-scope pass

## AI slop veto

Block on:
- any critical finding
- 2+ major findings
- 4+ minor findings
- cumulative pressure >= 6, minor=1 and major=3

## Revision routing

Return one `RevisionRequest` to the smallest responsible node:
- concept/originality -> `creative-director`
- insight depth -> `insight-mining`
- pattern saturation -> `campaign-canon`
- activation mechanic -> `brand-activation`
- narrative arc -> `visual-storytelling`
- strategy/message -> `creative-strategy`
- hierarchy/composition -> `composition-director`
- typography -> `typography-director`
- Arabic -> `arabic-rtl-director`
- brand/product identity -> `brand-intelligence`
- physical integration -> `manipulation-director`
- wrong annotation target -> `edit-sanitizer` with `annotation_mapping`
- over-broad edit scope -> `edit-sanitizer` with `edit_scope`
- collateral drift -> `edit-sanitizer` with `collateral_change`
- provider instruction mismatch -> `prompt-compiler`

---

## Cross-Skill Neural Connections & References

### Peer & Specialist Skills
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator and final signoff authority
- [Composition Director](../composition-director/SKILL.md) — Hierarchy & spatial layout revisions
- [Brand Intelligence](../brand-intelligence/SKILL.md) — Brand guidelines & product fidelity revisions
- [Edit Sanitizer](../edit-sanitizer/SKILL.md) — Bounded edit scope & annotation corrections
- [Prompt Compiler](../prompt-compiler/SKILL.md) — Provider prompt adjustments

### Schemas & References
- [Visual Review Schema](schemas/visual-review.schema.json) — Local review schema
- [Revision Request](../../shared/contracts/revision-request.schema.json) — Mesh revision contract
- [Edit Contract](../../shared/contracts/edit-contract.schema.json) — Bounded edit contract
- [Visual QA & Revisions](../../shared/references/visual-qa-and-revisions.md) — Review methodology
- [AI Slop Taxonomy](../../shared/references/ai-slop-taxonomy.md) — Slop families & severity criteria
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — 7-point representation checklist
- [Software Architecture](../../shared/references/software-architecture-and-contracts.md) — Bounded contexts & lock hierarchy

Referenced files: 7

View saved version →

---
name: visual-storytelling
description: Narrative architecture and visual storytelling specialist. This skill should be used when structuring dramatic narrative arcs, crafting multi-frame storyboards, applying classic storytelling frameworks (Story Spine, Sparkline, Freytag's Pyramid, Monroe's Motivated Sequence, Pixar Rules, Hero's Journey), or aligning visual pacing with complex emotional tiers.
---

# Visual Storytelling

Visual Storytelling structures cinematic narrative arcs, storyboard progressions, and visual pacing to evoke precise emotional transformation in the viewer.

---

## 1. The 6 Narrative Frameworks

1. **Kenn Adams' Story Spine**:
   - *Once upon a time... Every day... Until one day... Because of that... Because of that... Until finally... And ever since then...*
2. **Nancy Duarte's Sparkline**:
   - Continuous contrast between **What Is** (current frustrating reality) and **What Could Be** (transformed future state), ending with a New Bliss.
3. **Freytag's Dramatic Pyramid**:
   - Exposition $\rightarrow$ Inciting Incident $\rightarrow$ Rising Action $\rightarrow$ Climax $\rightarrow$ Falling Action $\rightarrow$ Resolution.
4. **Monroe's Motivated Sequence**:
   - Attention $\rightarrow$ Need $\rightarrow$ Satisfaction $\rightarrow$ Visualization $\rightarrow$ Action.
5. **Pixar's Rules of Story**:
   - Character desire over comfort; admiring a character for trying rather than for succeeding; simplifying focus.
6. **The Hero's Journey (Campellian/Vogler)**:
   - Departure $\rightarrow$ Initiation $\rightarrow$ Return with the Elixir.

---

## 2. Emotional Specificity (Tier 1/2/3 Hierarchy)

Great storytelling avoids generic "positive/happy" states and targets distinct Tier 2 and Tier 3 emotional registers:

- **Tier 1 (Generic, score $\le 6$)**: Happy, sad, excited, angry.
- **Tier 2 (Specific, score 6–8)**: Nostalgic longing, defiant pride, quiet vindication, bittersweet relief.
- **Tier 3 (Complex / Dual, score 8–10)**: Bittersweet triumph, vulnerable defiance, ironic sincerity, anxious optimism.

---

## 3. Cross-Skill Neural Connections & References

### Peer & Downstream Skills
- [Insight Mining](../insight-mining/SKILL.md) — Upstream emotional tensions & human truth
- [Composition Director](../composition-director/SKILL.md) — Visual pacing, focal paths, and frame hierarchy
- [Campaign DNA](../campaign-dna/SKILL.md) — Multi-frame continuity and sequential narrative rules
- [Creative Director](../creative-director/SKILL.md) — High-concept ideation & Cannes scoring
- [Designly Director](../designly-director/SKILL.md) — Lead orchestrator

### Domain References & Contracts
- [Storytelling Frameworks Guide](references/storytelling-frameworks.md) — 6 implementation cards and storyboard patterns
- [Emotion Hierarchy](references/emotion-hierarchy.md) — Tier 1/2/3 emotional register taxonomy
- [Inclusive Representation](../../shared/references/inclusive-representation-and-ethics.md) — Anti-stereotype character design
- [Signal Packet Schema](../../shared/contracts/signal-packet.schema.json) — Neural Mesh signal handoff
- [Design Context Schema](../../shared/contracts/design-context.schema.json) — Shared state context

Referenced files: 6