tribunal-kit 8.0.0 → 9.0.0

package/.agent/skills/property-based-testing/SKILL.md

@@ -2,8 +2,8 @@
 name: property-based-testing
 description: Generative input invariant testing using fast-check (TS/JS) and hypothesis (Python) to uncover hidden edge cases and boundary failures.
 tools: Read, Grep, Glob, Edit, Write
-version: 3.0.0
-last-updated: 2026-08-05
+version: 4.0.0
+last-updated: 2026-09-07
 script: .agent/scripts/test_runner.js
 scripts-binding:
   - .agent/scripts/test_runner.js
@@ -24,6 +24,12 @@ Before writing property tests:
 2. Arbitrary Generator Scoping → Constrain generator bounds to domain validity (e.g. non-empty strings, positive integers)
 3. Shrinking & Reproducibility → Store seed values for failing test runs to reproduce minimal failing inputs
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Generative input invariant testing using fast-check (TS/JS) and hypothesis (Python) to uncover hidden edge cases and boundary failures..
+- **DO NOT activate when:** The task falls strictly outside property-based-testing domain or belongs to a different dedicated specialist.
+
 ## Fast-Check Arbitrary Generator & Vitest Invariant Test
 
 ```typescript
@@ -52,7 +58,27 @@ test('currency parser invariant: non-negative parsed numbers', () => {
 });
 ```
 
-## 🛑 Verification-Before-Completion (VBC) Protocol
+---
+
+## 🏛️ 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
+```
+✅ 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?
+```
 
-- Run minimum 100 iterations per property test run.
-- Confirm shrinking mechanism isolates minimal failing counter-example on assertion failure.
+### 🛑 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/python-patterns/SKILL.md

@@ -2,8 +2,8 @@
 name: python-patterns
 description: Python development principles and decision-making. Framework selection, async patterns, type hints, project structure. Teaches thinking, not copying.
 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:
   - python-pro
   - clean-code
@@ -25,6 +25,12 @@ Before architecting Python projects or selecting frameworks, you MUST inspect:
 2. Specific Exception Handling (Section 16) → Catch explicit exceptions (`ValueError`, `AppError`); ban bare `except:` or generic `except Exception:` swallowing
 3. Pathlib Modern API (Section 17) → Use `pathlib.Path` for cross-platform filesystem operations; ban legacy `os.path`
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Python development principles and decision-making. Framework selection, async patterns, type hints, project structure. Teaches thinking, not copying..
+- **DO NOT activate when:** The task falls strictly outside python-patterns domain or belongs to a different dedicated specialist.
+
 ## Hallucination Traps (Read First)
 
 - ❌ Using `dict` for structured data when a dataclass/Pydantic model exists -> ✅ Dicts have no type safety; use typed models
@@ -33,8 +39,6 @@ Before architecting Python projects or selecting frameworks, you MUST inspect:
 
 ---
 
-# Python Development Principles
-
 ---
 
 ## Framework Selection
@@ -239,69 +243,17 @@ 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?
@@ -310,8 +262,6 @@ 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/python-pro/SKILL.md

@@ -2,8 +2,8 @@
 name: python-pro
 description: Python 3.12+ specialist. FastAPI, Pydantic v2, asyncio, modern types, pytest. Use when building Python APIs, data pipelines, automation, or any Python code.
 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:
   - python-patterns
   - data-validation-schemas
@@ -21,22 +21,44 @@ scripts-binding:
 
 Before writing Python code or FastAPI endpoints, you MUST inspect:
 
-1. Pydantic v2 Migration Syntax (Section 17) → Use `model_dump()`, `model_dump_json()`, `model_validate()`, and `@field_validator`; ban v1 `.dict()` and `@validator`
-2. Modern Native Generics (Section 16) → Use native `list[str]`, `dict[k,v]`, and `X | None` (Python 3.10+); ban legacy `typing.List` / `typing.Optional`
-3. Non-Blocking Async Clients (Section 21) → Use `httpx.AsyncClient()` in async code; ban blocking `requests` library inside asyncio event loops
-
-# Python 3.12+ — Dense Reference
+1. PEP 695 Syntax (Python 3.12+) → Use `type Alias = ...` and generic functions `def func[T](...)`; ban legacy `TypeVar` and `typing.Union`
+2. Structured Concurrency (`asyncio.TaskGroup`) → Use `async with asyncio.TaskGroup() as tg:` for parallel task execution; avoid raw `asyncio.gather()` error leaks
+3. Pydantic v2 Validation → Use `model_dump()`, `model_dump_json()`, `model_validate()`, and `@field_validator`; ban v1 `.dict()` and `@validator`
+4. Non-Blocking Async Clients → Use `httpx.AsyncClient()` in async code; ban blocking `requests` library inside asyncio event loops
+
+## Activation Boundaries
+
+- **Activate when:** Building Python 3.12+ APIs (FastAPI, Litestar, Django), asyncio pipelines, Pydantic v2 schemas, data processing, and pytest suites.
+- **DO NOT activate when:** Writing frontend TypeScript/JavaScript, or low-level C++/Rust code.
+
+## 2026 Python 3.12+ Performance & Logic Invariants
+
+1. **Structured Concurrency with `TaskGroup`**:
+   ```python
+   async with asyncio.TaskGroup() as tg:
+       task1 = tg.create_task(fetch_user(user_id))
+       task2 = tg.create_task(fetch_orders(user_id))
+   # If either fails, the other is cancelled immediately; exceptions grouped in ExceptionGroup
+   ```
+2. **PEP 695 Native Type Aliases & Generics**:
+   ```python
+   type Coordinates = tuple[float, float]
+   def get_first[T](items: list[T]) -> T | None:
+       return items[0] if items else None
+   ```
+3. **Pydantic v2 Zero-Copy Deserialization**: Use `TypeAdapter(list[Model]).validate_python(data)` for batch parsing.
+4. **Connection Pooling & Lifespan**: Always manage client lifecycles using `lifespan` context managers rather than re-instantiating HTTP/DB clients per request.
 
 ## Hallucination Traps (Read First)
 
-- ❌ `from typing import List, Dict, Optional, Union` → ✅ `list[str]`, `dict[k,v]`, `X | None`, `X | Y` (Python 3.10+)
+- ❌ `from typing import List, Dict, Optional, Union, TypeVar` → ✅ Native `list[str]`, `dict[k,v]`, `X | None`, `type Alias = ...`
 - ❌ `user.dict()` / `user.json()` / `UserCreate.parse_obj()` → ✅ Pydantic v2: `model_dump()`, `model_dump_json()`, `model_validate()`
 - ❌ Pydantic `class Config: orm_mode = True` → ✅ `model_config = {"from_attributes": True}`
 - ❌ `@validator` / `@root_validator` → ✅ `@field_validator` / `@model_validator`
-- ❌ `@app.on_event("startup")` → ✅ `lifespan` context manager (deprecated)
-- ❌ `import requests` in async code → ✅ `httpx.AsyncClient()` (requests BLOCKS the event loop)
-- ❌ `asyncio.run()` inside running loop → ✅ `await` directly or use `loop.create_task()`
-- ❌ `except Exception as e: pass` → ✅ always log or re-raise
+- ❌ `@app.on_event("startup")` → ✅ FastAPI `lifespan` context manager
+- ❌ `import requests` in async code → ✅ `httpx.AsyncClient()`
+- ❌ `asyncio.gather()` leaving orphaned tasks on error → ✅ `asyncio.TaskGroup()`
+- ❌ `except Exception as e: pass` → ✅ Always log or re-raise
 
 ---
 
@@ -304,69 +326,17 @@ 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?
@@ -375,8 +345,6 @@ 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/quieter/SKILL.md

@@ -1,8 +1,8 @@
 ---
 name: quieter
 description: Tone down overly loud, noisy, visually aggressive, or distracting designs while maintaining high visual quality. Use when a UI feels cluttered, overwhelming, tacky, or visually hyperactive.
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - distill
   - swiss-design
@@ -27,6 +27,12 @@ Before subduing visual design elements, you MUST inspect:
 
 Reduce visual noise, eliminate unnecessary color competition, and restore calm focus to chaotic interfaces.
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Tone down overly loud, noisy, visually aggressive, or distracting designs while maintaining high visual quality. Use when a UI feels cluttered, overwhelming, tacky, or visually hyperactive..
+- **DO NOT activate when:** The task falls strictly outside quieter domain or belongs to a different dedicated specialist.
+
 ---
 
 ## 5 Restraint Tactics
@@ -56,28 +62,25 @@ Reduce visual noise, eliminate unnecessary color competition, and restore calm f
 
 ---
 
-## 🤖 LLM-Specific Traps
+## 🏛️ Tribunal Verification & Guardrails
 
-1. **Making Text Unreadable**: Reducing color saturation so much that text fails WCAG AA contrast (below 4.5:1).
-2. **Stripping Functional Indicators**: Removing error/warning colors when reducing palette noise. Keep functional colors crisp.
-
----
+**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 saturated colors restricted to primary actions?
-✅ Did I retain minimum 4.5:1 text contrast for all body and label copy?
-✅ Has unnecessary layout animation or pulsing been eliminated?
+✅ 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/react-doctor/SKILL.md

@@ -1,8 +1,8 @@
 ---
 name: react-doctor
 description: Scan React and Next.js applications for security, performance, re-render inefficiencies, memory leaks, and correctness issues.
-version: 3.0.0
-last-updated: 2026-07-30
+version: 4.0.0
+last-updated: 2026-09-07
 skills:
   - react-specialist
   - nextjs-react-expert
@@ -27,6 +27,12 @@ Before auditing React components for performance or health issues, you MUST insp
 
 Diagnose and resolve unnecessary re-renders, state synchronization bugs, memory leaks, and hook dependency issues in React applications.
 
+
+## Activation Boundaries
+
+- **Activate when:** Operating in tasks requiring Scan React and Next.js applications for security, performance, re-render inefficiencies, memory leaks, and correctness issues..
+- **DO NOT activate when:** The task falls strictly outside react-doctor domain or belongs to a different dedicated specialist.
+
 ---
 
 ## 4 React Health Checks
@@ -58,28 +64,25 @@ useEffect(() => {
 
 ---
 
-## 🤖 LLM-Specific Traps
+## 🏛️ Tribunal Verification & Guardrails
 
-1. **Over-using `useCallback` everywhere**: Wrapping trivial primitives or 1-line functions that aren't passed to memoized children.
-2. **Missing `useEffect` Dependencies**: Omitting referenced variables from effect dependency arrays without proper `useCallback` or ref wrapping.
-
----
+**Slash command: `/review` or `/tribunal-full`**
+**Active reviewers: `logic-reviewer` · `security-auditor`**
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
-
-**Active reviewers: `react-specialist` · `type-safety`**
+### ❌ 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 event listeners and timers properly cleaned up in `useEffect` returns?
-✅ Are expensive calculations wrapped in `useMemo`?
-✅ Is state colocated to minimize re-render scope?
+✅ 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/react-specialist/SKILL.md

@@ -2,8 +2,8 @@
 name: react-specialist
 description: React 19+ specialist. use(), useActionState, useOptimistic, React Compiler, Server/Client Components, Zustand/Jotai, React Query. Use when building components, managing state, optimizing renders.
 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:
   - nextjs-react-expert
   - react-doctor
@@ -21,18 +21,33 @@ scripts-binding:
 
 Before writing React 19+ components or hooks, you MUST inspect:
 
-1. React 19 APIs (`useActionState` vs deprecated `useFormState`, `use()` vs `useContext`, `ref` as prop vs deprecated `forwardRef`)
-2. React Compiler Memoization (Section 125) → Avoid manual `useMemo`/`useCallback` unless React Compiler is explicitly disabled
-3. State Destructuring Traps (Section 189) → Never destructure entire Zustand stores; use granular selectors `useStore(s => s.value)`
+1. React 19 APIs (`useActionState` vs deprecated `useFormState`, `use()` vs `useContext`, `ref` as standard prop vs deprecated `forwardRef`)
+2. React Compiler Optimization → Rely on automatic memoization; avoid manual `useMemo`/`useCallback` unless compiling without React Compiler
+3. Granular Store Selectors → Never destructure entire stores (`const { a, b } = useStore()`); use atomic selectors (`useStore(s => s.a)`)
+4. Server/Client Serialization Boundary → Ensure data passed from RSC to `"use client"` components is JSON-serializable (no classes, functions, or Symbols)
+
+## Activation Boundaries
+
+- **Activate when:** Building React 19 web interfaces, custom hooks, Client/Server Components, interactive forms, optimistic UI, and render performance tuning.
+- **DO NOT activate when:** Writing pure backend Node.js APIs without UI, or styling pure CSS without React component interactions.
+
+## 2026 React 19 Performance & Architecture Invariants
+
+1. **Direct `ref` as a Prop**: Never use `forwardRef`. In React 19, `ref` is passed directly as a standard component prop.
+2. **Server Actions & `useActionState`**: Prefer native form actions and `useActionState` over manual `onSubmit` event handlers with `preventDefault()`.
+3. **Optimistic Updates**: Use `useOptimistic` for instant local feedback before network requests settle.
+4. **Transition-Aware State Updates**: Wrap non-urgent state updates in `startTransition` to keep main thread interactions responsive.
+5. **No Context Provider Wrapper Boilerplate**: In React 19, `<ThemeContext value={theme}>` replaces `<ThemeContext.Provider value={theme}>`.
 
 ## Hallucination Traps (Read First)
 
-- ❌ `useFormState` → ✅ `useActionState` (stable name)
-- ❌ `useContext()` in conditionals → ✅ `use(Context)` CAN be conditional
-- ❌ `useMemo/useCallback/React.memo` in React 19+ projects → ✅ React Compiler handles this
-- ❌ `exitBeforeEnter` (Framer) → ✅ `mode="wait"` on `<AnimatePresence>`
+- ❌ `forwardRef((props, ref) => ...)` → ✅ In React 19, accept `ref` directly as a component prop
+- ❌ `<ThemeContext.Provider value={...}>` → ✅ In React 19, use `<ThemeContext value={...}>`
+- ❌ `useFormState()` → ✅ `useActionState()` (stable API)
+- ❌ `useContext()` inside `if` conditionals → ✅ `use(Context)` CAN be called conditionally
+- ❌ Manual `useMemo`/`useCallback` everywhere → ✅ Let React Compiler handle memoization automatically
+- ❌ Destructuring whole store `const { user } = useStore()` → ✅ Granular selector `useStore(s => s.user)`
 - ❌ `next/router` → ✅ `next/navigation` in App Router
-- ❌ Server Components using `useState/useEffect` → must be `"use client"`
 
 ---
 
@@ -286,69 +301,17 @@ 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?
@@ -357,8 +320,6 @@ 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.