tribunal-kit 7.0.0 → 9.0.0

package/.agent/skills/trend-researcher/SKILL.md

@@ -2,8 +2,8 @@
 name: trend-researcher
 description: Creative muse and design trend analyzer for modern web/mobile interfaces.
 tools: Read, Grep, Glob, Bash, Edit, Write
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - frontend-design
   - web-design-guidelines
@@ -25,10 +25,14 @@ Before recommending design trends or visual directions, you MUST inspect:
 2. WCAG 2.2 AA Contrast Compliance (Section 61) → Calculate HSL palette contrast ($\ge 4.5:1$ text, $\ge 3:1$ large text) before finalizing tokens
 3. Verified Google Fonts Only (Section 175) → Ensure typography recommendations are active in Google Fonts or system defaults
 
-# Trend Researcher Skill
-
 You are a creative muse and design trend analyst. Your purpose is to bridge the gap between "generic UI" and "stunning, intentional design" by grounding aesthetic decisions in real-world trends, documented design systems, and measurable user impact.
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Creative muse and design trend analyzer for modern web/mobile interfaces..
+- **DO NOT activate when:** The task falls strictly outside trend-researcher domain or belongs to a different dedicated specialist.
+
 ## When to Activate
 
 - When asked to "brainstorm for design", "find inspiration", or "select a style guide".
@@ -219,69 +223,17 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-**Slash command: `/review` or `/tribunal-full`**
-**Active reviewers: `logic-reviewer` · `security-auditor`**
-
-### ❌ Forbidden AI Tropes
-
-1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
-2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
-3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
-
-Review these questions before confirming output:
-
-```
-✅ Did I rely ONLY on real, verified tools and methods?
-✅ Is this solution appropriately scoped to the user's constraints?
-✅ Did I handle potential failure modes and edge cases?
-✅ Have I avoided generic boilerplate that doesn't add value?
-```
-
-### 🛑 Verification-Before-Completion (VBC) Protocol
-
-**CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
-
-- ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
-- ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.
-
-## Pre-Flight Checklist
-
-- [ ] Have I reviewed the user's specific constraints and requests?
-- [ ] Have I checked the environment for relevant existing implementations?
-
-## VBC Protocol (Verification-Before-Completion)
-
-You MUST verify existing code signatures and variables before attempting to modify or call them. No hallucination is permitted.
-
----
-
-## 🤖 LLM-Specific Traps
-
-AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
-
-1. **Over-engineering:** Proposing complex abstractions or distributed systems when a simpler approach suffices.
-2. **Hallucinated Libraries/Methods:** Using non-existent methods or packages. Always `// VERIFY` or check `package.json` / `requirements.txt`.
-3. **Skipping Edge Cases:** Writing the "happy path" and ignoring error handling, timeouts, or data validation.
-4. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
-5. **Silent Degradation:** Catching and suppressing errors without logging or re-raising.
-
----
-
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+## 🏛️ Tribunal Verification & Guardrails
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
 
 ### ❌ Forbidden AI Tropes
-
 1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
 2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
 3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
 
 ### ✅ Pre-Flight Self-Audit
-
-Review these questions before confirming output:
-
 ```
 ✅ Did I rely ONLY on real, verified tools and methods?
 ✅ Is this solution appropriately scoped to the user's constraints?
@@ -290,8 +242,6 @@ Review these questions before confirming output:
 ```
 
 ### 🛑 Verification-Before-Completion (VBC) Protocol
-
 **CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
-
 - ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
 - ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.

package/.agent/skills/typescript-advanced/SKILL.md

@@ -2,8 +2,8 @@
 name: typescript-advanced
 description: Advanced TypeScript mastery. Generics with constraints, conditional types, mapped types, template literal types, the satisfies operator, discriminated unions, branded/nominal types, type-level programming, utility type internals, variance annotations, module augmentation, and declaration merging. Use when writing complex type definitions, building type-safe libraries, or solving "how do I type this?" problems.
 tools: Read, Grep, Glob, Bash, Edit, Write
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - clean-code
   - data-validation-schemas
@@ -21,19 +21,30 @@ scripts-binding:
 
 Before writing complex TypeScript types or library abstractions, you MUST inspect:
 
-1. Zero `as any` Cast Rule (Section 15) → Fix the underlying type signature or use `as unknown as T` with explicit reasoning comment; ban `as any`
+1. Zero `as any` Cast Rule (Section 15) → Fix the underlying type signature or use `unknown` with a type guard; ban `as any`
 2. Discriminated Unions for State Machine Modeling (Section 62) → Model multi-state objects with tagged discriminated unions for compile-time exhaustive checks
 3. The `satisfies` Operator vs Type Assertion (Section 193) → Use `satisfies` to validate types without widening object literals
+4. Compiler Memory & Type Instantiation Limits → Avoid recursive conditional types deeper than 5 levels; prefer interface extension over deep `&` intersections
+5. TS 5.5+ Inferred Type Predicates → Leverage native type narrowing in `.filter()` without manual `x is T` boilerplate where possible
 
-## Hallucination Traps (Read First)
+## Activation Boundaries
 
-- ❌ Using `as any` to silence type errors -> ✅ Fix the type; `as any` hides bugs that crash at runtime
-- ❌ Using `interface` when `type` is needed (unions, mapped types) -> ✅ `type` for unions/intersections/mapped; `interface` for objects that may be extended
-- ❌ Overcomplicating types — if a type takes 30 seconds to read, simplify it -> ✅ Types serve the developer, not the other way around
+- **Activate when:** Designing type-safe library APIs, generic utility types, discriminated union state machines, and resolving complex TypeScript compiler errors.
+- **DO NOT activate when:** Writing pure runtime JavaScript without TypeScript, or basic application components with trivial props.
 
----
+## 2026 TypeScript Performance & Compiler Invariants
 
-# Advanced TypeScript — Type-Level Mastery
+1. **Explicit Return Types on Exports (`isolatedDeclarations`)**: Always add explicit return types to exported functions/methods for fast, parallel build compilation.
+2. **Interface Extension Over Deep Intersections**: Use `interface B extends A` instead of `type B = A & { ... }`. Interfaces are cached by TS compiler's internal type-checker, preventing quadratic build slowdowns.
+3. **Safe Indexed Access**: Handle `undefined` when reading objects/arrays under `noUncheckedIndexedAccess`.
+4. **Const Type Parameters**: Use `function parse<const T>(val: T)` to preserve literal types without requiring the caller to write `as const`.
+
+## Hallucination Traps (Read First)
+
+- ❌ Using `as any` to silence type errors -> ✅ Fix the type or use `unknown` + type guard; `as any` masks runtime errors
+- ❌ Using deep recursive conditional types that trigger `Type instantiation is excessively deep` -> ✅ Use iteration or flat lookup tables
+- ❌ Overusing `type X = A & B & C & D` -> ✅ Use `interface` extension to preserve compiler performance
+- ❌ Manual `x is T` when TS 5.5+ infers the predicate -> ✅ Write natural predicate functions without unnecessary type assertion casts
 
 ---
 
@@ -319,69 +330,17 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-**Slash command: `/review` or `/tribunal-full`**
-**Active reviewers: `logic-reviewer` · `security-auditor`**
-
-### ❌ Forbidden AI Tropes
-
-1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
-2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
-3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
-
-Review these questions before confirming output:
-
-```
-✅ Did I rely ONLY on real, verified tools and methods?
-✅ Is this solution appropriately scoped to the user's constraints?
-✅ Did I handle potential failure modes and edge cases?
-✅ Have I avoided generic boilerplate that doesn't add value?
-```
-
-### 🛑 Verification-Before-Completion (VBC) Protocol
-
-**CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
-
-- ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
-- ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.
-
-## Pre-Flight Checklist
-
-- [ ] Have I reviewed the user's specific constraints and requests?
-- [ ] Have I checked the environment for relevant existing implementations?
-
-## VBC Protocol (Verification-Before-Completion)
-
-You MUST verify existing code signatures and variables before attempting to modify or call them. No hallucination is permitted.
-
----
-
-## 🤖 LLM-Specific Traps
-
-AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
-
-1. **Over-engineering:** Proposing complex abstractions or distributed systems when a simpler approach suffices.
-2. **Hallucinated Libraries/Methods:** Using non-existent methods or packages. Always `// VERIFY` or check `package.json` / `requirements.txt`.
-3. **Skipping Edge Cases:** Writing the "happy path" and ignoring error handling, timeouts, or data validation.
-4. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
-5. **Silent Degradation:** Catching and suppressing errors without logging or re-raising.
-
----
-
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+## 🏛️ Tribunal Verification & Guardrails
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
 
 ### ❌ Forbidden AI Tropes
-
 1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
 2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
 3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
 
 ### ✅ Pre-Flight Self-Audit
-
-Review these questions before confirming output:
-
 ```
 ✅ Did I rely ONLY on real, verified tools and methods?
 ✅ Is this solution appropriately scoped to the user's constraints?
@@ -390,8 +349,6 @@ Review these questions before confirming output:
 ```
 
 ### 🛑 Verification-Before-Completion (VBC) Protocol
-
 **CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
-
 - ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
 - ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.

package/.agent/skills/typeset/SKILL.md

@@ -1,8 +1,8 @@
 ---
 name: typeset
 description: Professional web typography scaling, font pairing, optical sizing, tracking, line height cadence, text-wrap balance, and OpenType features. Use when refining typography systems or text styling.
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - swiss-design
   - better-ui
@@ -27,6 +27,12 @@ Before refining typography scales or text styles, you MUST inspect:
 
 Architect disciplined, optically balanced typography systems across web components.
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Professional web typography scaling, font pairing, optical sizing, tracking, line height cadence, text-wrap balance, and OpenType features. Use when refining typography systems or text styling..
+- **DO NOT activate when:** The task falls strictly outside typeset domain or belongs to a different dedicated specialist.
+
 ---
 
 ## 4 Typography Rules
@@ -67,28 +73,25 @@ $$\text{Tracking}(\text{px}) \propto -\log(\text{FontSize})$$
 
 ---
 
-## 🤖 LLM-Specific Traps
+## 🏛️ Tribunal Verification & Guardrails
 
-1. **Uniform Line Height**: Setting `line-height: 1.5` on large 4rem display headings, causing massive awkward vertical gaps.
-2. **Missing Text Wrap Rules**: Allowing headings to wrap into single orphaned words.
-
----
+**Slash command: `/review` or `/tribunal-full`**
+**Active reviewers: `logic-reviewer` · `security-auditor`**
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
-
-**Active reviewers: `frontend-reviewer` · `ui-ux-auditor`**
+### ❌ Forbidden AI Tropes
+1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
+2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
+3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
 
 ### ✅ Pre-Flight Self-Audit
-
 ```
-✅ Are line-heights inversely proportional to font sizes (Display: 1.1, Body: 1.5)?
-✅ Is `text-wrap: balance` applied to display titles and headings?
-✅ Do numeric columns use tabular numbers (`font-variant-numeric: tabular-nums`)?
+✅ Did I rely ONLY on real, verified tools and methods?
+✅ Is this solution appropriately scoped to the user's constraints?
+✅ Did I handle potential failure modes and edge cases?
+✅ Have I avoided generic boilerplate that doesn't add value?
 ```
 
 ### 🛑 Verification-Before-Completion (VBC) Protocol
-
 **CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
-
 - ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
 - ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.

package/.agent/skills/ui-reasoning-engine/SKILL.md

@@ -1,8 +1,8 @@
 ---
 name: ui-reasoning-engine
 description: Enforces a structured cognitive pipeline that forces the agent to analyze user goals, information architecture, platform constraints, and design tokens before styling any component.
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - product-aware-heuristics
   - interface-design
@@ -28,6 +28,12 @@ Before generating UI component code, you MUST inspect:
 
 To build state-of-the-art interfaces, you must think before you style. This skill enforces a structured reasoning process. Before writing any HTML, CSS, or Component code, you must execute the following 16-step analysis.
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Enforces a structured cognitive pipeline that forces the agent to analyze user goals, information architecture, platform constraints, and design tokens before styling any component..
+- **DO NOT activate when:** The task falls strictly outside ui-reasoning-engine domain or belongs to a different dedicated specialist.
+
 ---
 
 ## The 16-Step Reasoning Pipeline
@@ -162,12 +168,25 @@ When building any UI, the Maker Agent must document this reasoning loop before r
 
 ---
 
-## Pre-Flight Checklist
+## 🏛️ Tribunal Verification & Guardrails
 
-- [ ] Have I executed the 16-step UI Reasoning Loop?
-- [ ] Have I identified the target Product Category and Visual Direction?
-- [ ] Have I verified touch targets are 24px minimum (and 44px+ for coarse pointers)?
+**Slash command: `/review` or `/tribunal-full`**
+**Active reviewers: `logic-reviewer` · `security-auditor`**
 
-## VBC Protocol (Verification-Before-Completion)
+### ❌ Forbidden AI Tropes
+1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
+2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
+3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
+
+### ✅ Pre-Flight Self-Audit
+```
+✅ Did I rely ONLY on real, verified tools and methods?
+✅ Is this solution appropriately scoped to the user's constraints?
+✅ Did I handle potential failure modes and edge cases?
+✅ Have I avoided generic boilerplate that doesn't add value?
+```
 
-You MUST verify existing code signatures and variables before attempting to modify or call them. No hallucination is permitted.
+### 🛑 Verification-Before-Completion (VBC) Protocol
+**CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
+- ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
+- ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.

package/.agent/skills/ui-skill-packs/SKILL.md

@@ -1,8 +1,8 @@
 ---
 name: ui-skill-packs
 description: Mandatory skill loading packs for UI generation. Consolidates 42+ individual UI skills into 3 tiered packs (Core Craft, Immersive & System, High-End Design) to guarantee non-generic, high-fidelity UI outputs.
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - taste-skill
   - better-colors
@@ -27,6 +27,12 @@ Before generating UI code, you MUST inspect:
 2. `DESIGN.md` / color token definitions → Verify palette rules and dark mode contrast variables
 3. Active UI component target → Assign Pack 1 (Core UI), Pack 2 (System Micro-Interactions), or Pack 3 (Immersive Landing)
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Mandatory skill loading packs for UI generation. Consolidates 42+ individual UI skills into 3 tiered packs (Core Craft, Immersive & System, High-End Design) to guarantee non-generic, high-fidelity UI outputs..
+- **DO NOT activate when:** The task falls strictly outside ui-skill-packs domain or belongs to a different dedicated specialist.
+
 ## 1. Pack Definitions
 
 When generating UI code, the agent MUST load the appropriate UI Skill Pack based on the complexity level of the requested interface:
@@ -62,21 +68,25 @@ Binding skills: taste-skill, better-colors, micro-interaction, impeccable...
 
 ---
 
-## Pre-Flight Checklist
-
-Before generating UI component code:
+## 🏛️ Tribunal Verification & Guardrails
 
-1. Verify target framework (React, Next.js, Vue, Svelte) and styling system (CSS Modules, Tailwind, OKLCH CSS).
-2. Select appropriate UI Skill Pack (Pack 1, Pack 2, or Pack 3).
-3. Ensure color variables use OKLCH color space for accessible contrast ratios.
-4. Verify micro-interaction focus and hover states for keyboard and mouse accessibility.
+**Slash command: `/review` or `/tribunal-full`**
+**Active reviewers: `logic-reviewer` · `security-auditor`**
 
----
-
-## VBC Protocol
+### ❌ Forbidden AI Tropes
+1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
+2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
+3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
 
-Visual Balance & Contrast Protocol:
+### ✅ Pre-Flight Self-Audit
+```
+✅ Did I rely ONLY on real, verified tools and methods?
+✅ Is this solution appropriately scoped to the user's constraints?
+✅ Did I handle potential failure modes and edge cases?
+✅ Have I avoided generic boilerplate that doesn't add value?
+```
 
-1. **Contrast Ratio**: Enforce WCAG 2.2 AA (4.5:1 text, 3:1 graphical elements).
-2. **Typography Cadence**: Enforce strict fluid font sizing and line-height cadence.
-3. **Motion Hygiene**: Respect `prefers-reduced-motion` media queries on all animations.
+### 🛑 Verification-Before-Completion (VBC) Protocol
+**CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
+- ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
+- ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.

package/.agent/skills/ui-skills-root/SKILL.md

@@ -1,8 +1,8 @@
 ---
 name: ui-skills-root
 description: Master router skill for design engineering and UI craft. Use when the user needs UI help and you must route by topic, stack, intent, or design discipline to the smallest useful set of UI skills.
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - ui-skill-packs
   - ui-reasoning-engine
@@ -29,6 +29,12 @@ Before routing UI requests, you MUST inspect:
 
 The master decision engine for routing user UI requests to the precise set of design-engineering skills.
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Master router skill for design engineering and UI craft. Use when the user needs UI help and you must route by topic, stack, intent, or design discipline to the smallest useful set of UI skills..
+- **DO NOT activate when:** The task falls strictly outside ui-skills-root domain or belongs to a different dedicated specialist.
+
 ---
 
 ## Intent-Based Skill Routing Matrix
@@ -76,23 +82,25 @@ When a user asks for UI assistance, identify their core **intent** and dispatch
 
 ---
 
-## 🤖 LLM-Specific Traps
+## 🏛️ Tribunal Verification & Guardrails
 
-1. **Skill Bloat:** Loading more than 3 skills at once for a single request. Pick the 1-2 highest precision skills.
-2. **Ignoring Intent:** Applying animation skills when the user asked for simple layout structure.
+**Slash command: `/review` or `/tribunal-full`**
+**Active reviewers: `logic-reviewer` · `security-auditor`**
 
----
-
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+### ❌ Forbidden AI Tropes
+1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
+2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
+3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
 
 ### ✅ Pre-Flight Self-Audit
-
 ```
-✅ Did I select the most specific UI skill for the user's explicit intent?
-✅ Am I respecting modern web design standards (CSS Variables, container queries, OKLCH)?
-✅ Is the proposed UI accessible (WCAG 2.2 AA contrast, keyboard nav)?
+✅ Did I rely ONLY on real, verified tools and methods?
+✅ Is this solution appropriately scoped to the user's constraints?
+✅ Did I handle potential failure modes and edge cases?
+✅ Have I avoided generic boilerplate that doesn't add value?
 ```
 
 ### 🛑 Verification-Before-Completion (VBC) Protocol
-
-You MUST verify that selected UI skills are present in `.agent/skills/` and read their instructions before generating code.
+**CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
+- ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
+- ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.

package/.agent/skills/ui-ux-pro-max/SKILL.md

@@ -1,8 +1,8 @@
 ---
 name: ui-ux-pro-max
 description: The Picasso Protocol — Elite UI/UX design mastery. Integrates the 16-step UI Reasoning Engine, Category-Specific Heuristics, and Anti-Pattern controls to generate portfolio-grade interfaces.
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - ui-reasoning-engine
   - product-aware-heuristics
@@ -22,11 +22,24 @@ scripts-binding:
 
 Before generating UI designs or front-end components, you MUST inspect:
 
-1. Rejection Guardrails (Section 17) → Ban primary purple gradients (`from-purple-600`), Inter/Roboto defaults, left-text/right-image hero layouts, and mesh gradients
-2. OKLCH Display-P3 Color Science (Section 29) → Use OKLCH variables with chroma `0.20`–`0.28` for accents and stepped neutral contrast (`0.04` Lightness increments)
-3. Fluid Typographic Scaling (Section 51) → Apply `clamp()` sizes, tight negative tracking (`-0.04em`), and `text-wrap: balance` on headings
+1. Rejection Guardrails → Ban primary purple gradients (`from-purple-600`), Inter/Roboto defaults, left-text/right-image hero layouts, and mesh gradients
+2. OKLCH Display-P3 Color Science → Use OKLCH variables with chroma `0.20`–`0.28` for accents and stepped neutral contrast (`0.04` Lightness increments)
+3. Fluid Typographic Scaling → Apply `clamp()` sizes, tight negative tracking (`-0.04em`), and `text-wrap: balance` on headings
+4. APCA Contrast Compliance → Verify text against APCA contrast thresholds rather than outdated WCAG 2.1 math
 
-This skill transforms AI-generated interfaces from generic "AI templates" into portfolio-grade, human-crafted interfaces by running intent analysis _before_ layout and styling.
+## Activation Boundaries
+
+- **Activate when:** Creating showcase landing pages, web application dashboards, design systems, design tokens, visual redesigns, and portfolio-grade UI components.
+- **DO NOT activate when:** Writing raw backend SQL queries, server infrastructure Dockerfiles, or CLI terminal utilities.
+
+## 2026 Picasso Protocol Design Invariants
+
+1. **Anti-Cliché Palette Discipline**:
+   Never use purple (#7c3aed / #8b5cf6) as primary brand color. Curate characterful palettes: Obsidian + Signal Orange, Slate + Electric Cyan, or Bone White + Deep Emerald.
+2. **Intentional Typography Pairing**:
+   Pair an expressive display font (e.g., Clash Display, Syne, Cabinet Grotesk, Newsreader) with an ultra-readable neutral body font (e.g., Geist Sans, Plus Jakarta Sans, General Sans).
+3. **Fluid Layout Bounds & Asymmetry**:
+   Break repetitive bento grids. Use asymmetric column spans (e.g. 7:5 or 8:4), overlapping cards, and ambient grain overlays.
 
 ---
 
@@ -108,12 +121,25 @@ All buttons and active controls must define styled properties for:
 
 ---
 
-## Pre-Flight Checklist
+## 🏛️ Tribunal Verification & Guardrails
 
-- [ ] Have I reviewed the user's specific constraints and requests?
-- [ ] Have I checked the environment for relevant existing implementations?
-- [ ] Have I verified the design passes the Anti-AI-Slop Checklist?
+**Slash command: `/review` or `/tribunal-full`**
+**Active reviewers: `logic-reviewer` · `security-auditor`**
 
-## VBC Protocol (Verification-Before-Completion)
+### ❌ Forbidden AI Tropes
+1. **Blind Assumptions:** Never make an assumption without documenting it clearly with `// VERIFY: [reason]`.
+2. **Silent Degradation:** Catching and suppressing errors without logging or handling.
+3. **Context Amnesia:** Forgetting the user's constraints and offering generic advice instead of tailored solutions.
+
+### ✅ Pre-Flight Self-Audit
+```
+✅ Did I rely ONLY on real, verified tools and methods?
+✅ Is this solution appropriately scoped to the user's constraints?
+✅ Did I handle potential failure modes and edge cases?
+✅ Have I avoided generic boilerplate that doesn't add value?
+```
 
-You MUST verify existing code signatures and variables before attempting to modify or call them. No hallucination is permitted.
+### 🛑 Verification-Before-Completion (VBC) Protocol
+**CRITICAL:** You must follow a strict "evidence-based closeout" state machine.
+- ❌ **Forbidden:** Declaring a task complete because the output "looks correct."
+- ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended.