tribunal-kit 4.2.0 → 4.3.1

package/.agent/skills/nextjs-react-expert/SKILL.md

@@ -1,192 +1,192 @@
----
-name: nextjs-react-expert
-description: Next.js 15+ App Router mastery. Server Components, Server Actions, PPR, caching, metadata, middleware, parallel/intercepting routes. Use when building Next.js apps or optimizing Next.js performance.
-allowed-tools: Read, Write, Edit, Glob, Grep
-version: 3.1.0
-last-updated: 2026-04-06
-applies-to-model: gemini-3-1-pro, claude-3-7-sonnet
----
-
-# Next.js 15+ App Router — Dense Reference
-
-## Hallucination Traps (Read First)
-- ❌ `pages/api/` or `_app.tsx` → ✅ App Router only: `app/api/route.ts`, `app/layout.tsx`
-- ❌ `getServerSideProps` → ✅ `async function Page()` fetches directly
-- ❌ `next/router` → ✅ `next/navigation` (`useRouter`, `usePathname`, `useSearchParams`)
-- ❌ Server Action without `"use server"` → ✅ Required at top of file or top of function
-- ❌ Fetch is cached by default → ✅ **Next.js 15 changed this**: `fetch()` is UNCACHED by default
-- ❌ `cookies()` at top of page → ✅ Opts entire route into dynamic rendering, breaking PPR. Wrap inside `<Suspense>`
-- ❌ Passing functions as props Server → Client → ✅ Illegal. Use Server Actions instead.
-- ❌ Plain `Response` in route handler → ✅ Use `NextResponse.json()`
-
----
-
-## App Router Conventions
-
-```text
-app/
-├── layout.tsx         ← Root shell (HTML/BODY)
-├── page.tsx           ← Route UI
-├── loading.tsx        ← Auto-suspense fallback
-├── error.tsx          ← Error boundary (Must be "use client")
-├── not-found.tsx      ← 404 UI
-├── global-error.tsx   ← Root error boundary
-├── api/users/route.ts ← API Handler (GET, POST)
-├── @modal/login/page.tsx ← Parallel Route (renders in same layout)
-└── (auth)/login/page.tsx ← Route Group (doesn't affect URL)
-```
-
----
-
-## Server vs Client Components
-
-- **Server Components (Default)**: Zero JS. Direct DB access. Secure env vars. 
-- **Client Components (`"use client"`)**: Lifecycle (`useEffect`), State (`useState`), Browser APIs (`window`), Event listeners (`onClick`).
-
-```tsx
-// ✅ INTERLEAVING PATTERN: Pass Server Component as children to Client Component
-export default function Page() {
-  return (
-    <ClientSidebar> {/* "use client" */}
-      <ServerStats /> {/* Server: zero JS bundle, fetches DB */}
-    </ClientSidebar>
-  );
-}
-```
-
----
-
-## Server Actions (Mutations)
-
-```tsx
-"use server"
-import { revalidatePath } from "next/cache";
-import { z } from "zod";
-
-const Schema = z.object({ name: z.string().min(2) });
-
-export async function createUser(prevState: any, formData: FormData) {
-  // ❌ TRAP: ALWAYS validate formData. Never trust client input.
-  const parsed = Schema.safeParse({ name: formData.get("name") });
-  if (!parsed.success) return { errors: parsed.error.flatten().fieldErrors };
-  
-  await db.user.create({ data: parsed.data });
-  revalidatePath("/users"); // Clears cache so next render shows new user
-  return { success: true };
-}
-```
-
-Client usage (React 19):
-```tsx
-"use client"
-import { useActionState } from "react";
-import { createUser } from "./actions";
-
-export function UserForm() {
-  const [state, formAction, isPending] = useActionState(createUser, null);
-  return (
-    <form action={formAction}>
-      <input name="name" />
-      <button disabled={isPending}>Submit</button>
-      {state?.errors?.name && <p>{state.errors.name}</p>}
-    </form>
-  )
-}
-```
-
----
-
-## Data Fetching & Caching (Next.js 15)
-
-```tsx
-// Next.js 15 caching defaults
-const dynamic = await fetch(url);                                  // 15 default: NO CACHE
-const static = await fetch(url, { cache: "force-cache" });         // Static
-const isr = await fetch(url, { next: { revalidate: 3600 } });      // Revalidate every hour
-const tagged = await fetch(url, { next: { tags: ["user-1"] } });   // On-demand via revalidateTag()
-
-// DB calls without fetch
-import { unstable_cache } from "next/cache";
-const getCachedUser = unstable_cache(
-  async (id) => db.user.findUnique({ where: { id } }),
-  ["user-cache-key"],
-  { revalidate: 60, tags: ["users"] }
-);
-```
-
-### Waterfall Elimination
-```tsx
-// ✅ Parallel Fetching:
-const [user, posts] = await Promise.all([getUser(), getPosts()]);
-
-// ✅ Streaming (PPR-compatible):
-export default function Page() {
-  return (
-    <main>
-      <FastNav />
-      {/* Page shell loads instantly, SlowChart streams in when ready */}
-      <Suspense fallback={<Skeleton />}>
-        <SlowChart /> 
-      </Suspense>
-    </main>
-  );
-}
-```
-
----
-
-## Partial Prerendering (PPR)
-
-PPR static-generates the route shell and streams dynamic parts. 
-```tsx
-// next.config.ts
-export default { experimental: { ppr: true } };
-
-// Any component reading cookies/headers inside a Suspense boundary becomes a dynamic hole
-import { cookies } from "next/headers";
-
-async function Cart() {
-  const c = await cookies(); // Next.js 15 cookies are async!
-  const cartId = c.get("cartId");
-}
-
-export default function Page() {
-  return (
-    <div>
-      <StaticHeader /> {/* Cached at build time on CDN */}
-      <Suspense fallback={<CartSkeleton />}>
-        <Cart /> {/* Dynamic, streamed at request time */}
-      </Suspense>
-    </div>
-  )
-}
-```
-
----
-
-## Middleware
-
-```typescript
-// middleware.ts (Root of project)
-import { NextResponse } from "next/server";
-import type { NextRequest } from "next/server";
-
-export function middleware(req: NextRequest) {
-  const token = req.cookies.get("auth-token");
-  if (!token && req.nextUrl.pathname.startsWith("/dashboard")) {
-    return NextResponse.redirect(new URL("/login", req.url));
-  }
-}
-
-export const config = {
-  matcher: ["/dashboard/:path*"], // Strict matcher is critical for performance
-};
-```
+---
+name: nextjs-react-expert
+description: Next.js 15+ App Router mastery. Server Components, Server Actions, PPR, caching, metadata, middleware, parallel/intercepting routes. Use when building Next.js apps or optimizing Next.js performance.
+allowed-tools: Read, Write, Edit, Glob, Grep
+version: 3.1.0
+last-updated: 2026-04-06
+applies-to-model: gemini-3-1-pro, claude-3-7-sonnet
+---
+
+# Next.js 15+ App Router — Dense Reference
+
+## Hallucination Traps (Read First)
+- ❌ `pages/api/` or `_app.tsx` → ✅ App Router only: `app/api/route.ts`, `app/layout.tsx`
+- ❌ `getServerSideProps` → ✅ `async function Page()` fetches directly
+- ❌ `next/router` → ✅ `next/navigation` (`useRouter`, `usePathname`, `useSearchParams`)
+- ❌ Server Action without `"use server"` → ✅ Required at top of file or top of function
+- ❌ Fetch is cached by default → ✅ **Next.js 15 changed this**: `fetch()` is UNCACHED by default
+- ❌ `cookies()` at top of page → ✅ Opts entire route into dynamic rendering, breaking PPR. Wrap inside `<Suspense>`
+- ❌ Passing functions as props Server → Client → ✅ Illegal. Use Server Actions instead.
+- ❌ Plain `Response` in route handler → ✅ Use `NextResponse.json()`
+
+---
+
+## App Router Conventions
+
+```text
+app/
+├── layout.tsx         ← Root shell (HTML/BODY)
+├── page.tsx           ← Route UI
+├── loading.tsx        ← Auto-suspense fallback
+├── error.tsx          ← Error boundary (Must be "use client")
+├── not-found.tsx      ← 404 UI
+├── global-error.tsx   ← Root error boundary
+├── api/users/route.ts ← API Handler (GET, POST)
+├── @modal/login/page.tsx ← Parallel Route (renders in same layout)
+└── (auth)/login/page.tsx ← Route Group (doesn't affect URL)
+```
+
+---
+
+## Server vs Client Components
+
+- **Server Components (Default)**: Zero JS. Direct DB access. Secure env vars. 
+- **Client Components (`"use client"`)**: Lifecycle (`useEffect`), State (`useState`), Browser APIs (`window`), Event listeners (`onClick`).
+
+```tsx
+// ✅ INTERLEAVING PATTERN: Pass Server Component as children to Client Component
+export default function Page() {
+  return (
+    <ClientSidebar> {/* "use client" */}
+      <ServerStats /> {/* Server: zero JS bundle, fetches DB */}
+    </ClientSidebar>
+  );
+}
+```
+
+---
+
+## Server Actions (Mutations)
+
+```tsx
+"use server"
+import { revalidatePath } from "next/cache";
+import { z } from "zod";
+
+const Schema = z.object({ name: z.string().min(2) });
+
+export async function createUser(prevState: any, formData: FormData) {
+  // ❌ TRAP: ALWAYS validate formData. Never trust client input.
+  const parsed = Schema.safeParse({ name: formData.get("name") });
+  if (!parsed.success) return { errors: parsed.error.flatten().fieldErrors };
+  
+  await db.user.create({ data: parsed.data });
+  revalidatePath("/users"); // Clears cache so next render shows new user
+  return { success: true };
+}
+```
+
+Client usage (React 19):
+```tsx
+"use client"
+import { useActionState } from "react";
+import { createUser } from "./actions";
 
+export function UserForm() {
+  const [state, formAction, isPending] = useActionState(createUser, null);
+  return (
+    <form action={formAction}>
+      <input name="name" />
+      <button disabled={isPending}>Submit</button>
+      {state?.errors?.name && <p>{state.errors.name}</p>}
+    </form>
+  )
+}
+```
 
 ---
 
-## 🤖 LLM-Specific Traps
+## Data Fetching & Caching (Next.js 15)
+
+```tsx
+// Next.js 15 caching defaults
+const dynamic = await fetch(url);                                  // 15 default: NO CACHE
+const static = await fetch(url, { cache: "force-cache" });         // Static
+const isr = await fetch(url, { next: { revalidate: 3600 } });      // Revalidate every hour
+const tagged = await fetch(url, { next: { tags: ["user-1"] } });   // On-demand via revalidateTag()
+
+// DB calls without fetch
+import { unstable_cache } from "next/cache";
+const getCachedUser = unstable_cache(
+  async (id) => db.user.findUnique({ where: { id } }),
+  ["user-cache-key"],
+  { revalidate: 60, tags: ["users"] }
+);
+```
+
+### Waterfall Elimination
+```tsx
+// ✅ Parallel Fetching:
+const [user, posts] = await Promise.all([getUser(), getPosts()]);
+
+// ✅ Streaming (PPR-compatible):
+export default function Page() {
+  return (
+    <main>
+      <FastNav />
+      {/* Page shell loads instantly, SlowChart streams in when ready */}
+      <Suspense fallback={<Skeleton />}>
+        <SlowChart /> 
+      </Suspense>
+    </main>
+  );
+}
+```
+
+---
+
+## Partial Prerendering (PPR)
+
+PPR static-generates the route shell and streams dynamic parts. 
+```tsx
+// next.config.ts
+export default { experimental: { ppr: true } };
+
+// Any component reading cookies/headers inside a Suspense boundary becomes a dynamic hole
+import { cookies } from "next/headers";
+
+async function Cart() {
+  const c = await cookies(); // Next.js 15 cookies are async!
+  const cartId = c.get("cartId");
+}
+
+export default function Page() {
+  return (
+    <div>
+      <StaticHeader /> {/* Cached at build time on CDN */}
+      <Suspense fallback={<CartSkeleton />}>
+        <Cart /> {/* Dynamic, streamed at request time */}
+      </Suspense>
+    </div>
+  )
+}
+```
+
+---
+
+## Middleware
+
+```typescript
+// middleware.ts (Root of project)
+import { NextResponse } from "next/server";
+import type { NextRequest } from "next/server";
+
+export function middleware(req: NextRequest) {
+  const token = req.cookies.get("auth-token");
+  if (!token && req.nextUrl.pathname.startsWith("/dashboard")) {
+    return NextResponse.redirect(new URL("/login", req.url));
+  }
+}
+
+export const config = {
+  matcher: ["/dashboard/:path*"], // Strict matcher is critical for performance
+};
+```
+
+
+---
+
+
 
 AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
 
@@ -198,7 +198,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
@@ -209,7 +209,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 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:
 ```
@@ -223,4 +223,12 @@ Review these questions before confirming output:
 
 **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.
+- ✅ **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.

package/.agent/skills/nodejs-best-practices/SKILL.md

@@ -500,7 +500,7 @@ if (isMainThread) {
 
 ---
 
-## 🤖 LLM-Specific Traps
+
 
 AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
 
@@ -512,7 +512,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
@@ -523,7 +523,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 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:
 ```
@@ -537,4 +537,12 @@ Review these questions before confirming output:
 
 **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.
+- ✅ **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.

package/.agent/skills/observability/SKILL.md

@@ -295,7 +295,7 @@ Alert design rules:
 
 ---
 
-## 🤖 LLM-Specific Traps
+
 
 AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
 
@@ -307,7 +307,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
@@ -318,7 +318,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 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:
 ```
@@ -332,4 +332,12 @@ Review these questions before confirming output:
 
 **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.
+- ✅ **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.

package/.agent/skills/parallel-agents/SKILL.md

@@ -98,7 +98,7 @@ for (const result of results) {
 
 ---
 
-## 🤖 LLM-Specific Traps
+
 
 AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
 
@@ -110,7 +110,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
@@ -121,7 +121,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 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:
 ```
@@ -135,4 +135,12 @@ Review these questions before confirming output:
 
 **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.
+- ✅ **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.

package/.agent/skills/performance-profiling/SKILL.md

@@ -219,7 +219,7 @@ Performance budget targets:
 
 ---
 
-## 🤖 LLM-Specific Traps
+
 
 AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
 
@@ -231,7 +231,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
@@ -242,7 +242,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 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:
 ```
@@ -256,4 +256,12 @@ Review these questions before confirming output:
 
 **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.
+- ✅ **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.

package/.agent/skills/plan-writing/SKILL.md

@@ -94,7 +94,7 @@ Unlike the high-level `implementation_plan.md`, the `task.md` serves as the live
 
 ---
 
-## 🤖 LLM-Specific Traps
+
 
 AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
 
@@ -106,7 +106,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
@@ -117,7 +117,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 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:
 ```
@@ -131,4 +131,12 @@ Review these questions before confirming output:
 
 **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.
+- ✅ **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.

package/.agent/skills/platform-engineer/SKILL.md

@@ -99,7 +99,7 @@ Ensure your infrastructure proposals abstract away the YAML mechanics. Give the
 
 ---
 
-## 🤖 LLM-Specific Traps
+
 
 AI coding assistants often fall into specific bad habits when dealing with this domain. These are strictly forbidden:
 
@@ -111,7 +111,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 
 ---
 
-## 🏛️ Tribunal Integration (Anti-Hallucination)
+
 
 **Slash command: `/review` or `/tribunal-full`**
 **Active reviewers: `logic-reviewer` · `security-auditor`**
@@ -122,7 +122,7 @@ AI coding assistants often fall into specific bad habits when dealing with this
 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:
 ```
@@ -136,4 +136,12 @@ Review these questions before confirming output:
 
 **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.
+- ✅ **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.