agent-enderun 1.1.5 → 1.1.7

package/templates/standards/quality-standards.md

@@ -0,0 +1,31 @@
+# ⚖️ Kurumsal Kod Kalitesi ve Disiplini (Linting & Standards)
+
+Bu belge, Agent Enderun ordusu tarafından üretilen tüm kodların uyması gereken teknik disiplin kurallarını tanımlar. Amaç; sıfır hata, maksimum okunabilirlik ve sürdürülebilirliktir.
+
+## 1. TypeScript Disiplini
+- **Strict Mode:** Tüm projelerde `strict: true` zorunludur.
+- **No Explicit Any:** `any` kullanımı kesinlikle yasaktır. Belirsiz tipler için `unknown` kullanılmalı ve tip muhafazaları (type guards) ile doğrulanmalıdır.
+- **Exhaustive Checks:** `switch-case` yapılarında tüm durumlar (exhaustive) kontrol edilmeli veya `never` tipi ile güvenlik sağlanmalıdır.
+
+## 2. ESLint ve Statik Analiz
+- **Zero Warnings:** Kod tabanında hiçbir ESLint uyarısı (warning) veya hatası (error) bulunamaz.
+- **Naming Conventions:**
+  - Değişkenler ve fonksiyonlar: `camelCase`
+  - Sınıflar ve Tipler: `PascalCase`
+  - Sabitler: `UPPER_SNAKE_CASE`
+  - Dosya isimleri: `kebab-case`
+- **Imports:** Kullanılmayan importlar otomatik olarak temizlenmeli ve import sırası düzenli olmalıdır.
+
+## 3. Formatlama (Prettier)
+- **Consistency:** Tüm dosyalarda tek tip formatlama zorunludur.
+- **Indentation:** 4 boşluk (4 spaces).
+- **Quotes:** Çift tırnak (`"`) kullanımı.
+- **Semicolons:** Her ifade noktalı virgül (`;`) ile bitmelidir.
+
+## 4. Hata Yönetimi (Error Handling)
+- **No Silent Failures:** Boş `catch` blokları yasaktır. Her hata en azından loglanmalı veya üst katmana iletilmelidir.
+- **Custom Errors:** İş mantığı hataları için projeye özgü `AppError` sınıfları kullanılmalıdır.
+
+## 5. Denetim Döngüsü
+- **Pre-commit:** Kod commit edilmeden önce mutlaka `npm run lint` ve `npm run type-check` çalıştırılmalıdır.
+- **@quality Audit:** Tüm değişiklikler `@quality` ajanı tarafından bu standartlara göre denetlenir.

package/templates/standards/security-standards.md

@@ -0,0 +1,21 @@
+# 🛡️ Kurumsal Güvenlik ve Veri Koruma Standartları
+
+Bu belge, Agent Enderun tarafından yönetilen projelerdeki güvenlik protokollerini ve veri koruma standartlarını tanımlar.
+
+## 1. Veri Tabanı Güvenliği (RLS)
+- **Row Level Security (RLS):** Tüm kullanıcı verileri RLS politikaları ile korunmalıdır. Bir kullanıcı, `@manager` tarafından onaylanmış bir izin protokolü dışında başka bir kullanıcının verisine erişemez.
+- **SQL Injection:** Ham SQL sorguları kesinlikle yasaktır (Kysely zorunluluğu).
+- **Encryption at Rest:** Hassas veriler (PII) veritabanında şifrelenmiş olarak saklanmalıdır.
+
+## 2. Kimlik Doğrulama ve Yetkilendirme
+- **JWT / Session:** Sadece `@security` ajanı tarafından onaylanmış güvenli kimlik doğrulama yöntemleri kullanılır.
+- **Role Based Access Control (RBAC):** Yetkilendirme şemaları `apps/backend/src/security/roles.ts` altında merkezi olarak yönetilmelidir.
+
+## 3. API Güvenliği
+- **CORS:** Sadece izin verilen domainler API'ye erişebilir.
+- **Rate Limiting:** Tüm API uç noktaları için rate limiting uygulanmalıdır.
+- **Input Validation:** Zod veya benzeri bir kütüphane ile tüm girdiler (request body, params) sıkı bir şekilde doğrulanmalıdır.
+
+## 4. Sır Yönetimi (Secret Management)
+- **No Hardcoded Secrets:** API anahtarları, şifreler veya gizli anahtarlar asla kod içinde saklanamaz.
+- **.env Discipline:** Tüm gizli değişkenler `.env` dosyası üzerinden yönetilmeli ve bu dosya asla git'e pushlanmamalıdır.

package/templates/standards/tailwind-standards.md

@@ -0,0 +1,20 @@
+# 🌊 Kurumsal Tailwind CSS Standartları
+
+Bu belge, Tailwind CSS kullanılan projelerde kod kalitesini ve tasarım disiplinini korumak için gereken kuralları tanımlar.
+
+## 1. Tasarım Disiplini ve Kısıtlamalar
+- **No Arbitrary Values:** `h-[123px]` veya `bg-[#fafafa]` gibi rastgele değerler kullanılması yasaktır. Tüm değerler `tailwind.config.ts` içindeki temadan beslenmelidir.
+- **Utility-First, Not Utility-Only:** Çok karmaşıklaşan class yapıları `@apply` ile veya bileşenlere bölünerek (Atomic Design) yönetilmelidir.
+- **Prettier Plugin:** Sınıf sıralaması için mutlaka `prettier-plugin-tailwindcss` kullanılmalıdır.
+
+## 2. Responsive ve Mobil-First
+- **Mobile-First:** Stiller önce mobil için yazılmalı, ardından `sm:`, `md:`, `lg:` prefixleri ile genişletilmelidir.
+- **Consistency:** Proje genelinde tutarlı breakpoint kullanımı (standard Tailwind breakpoints) zorunludur.
+
+## 3. Temiz Kod ve Organizasyon
+- **Clean Templates:** HTML/JSX içinde 10'dan fazla Tailwind sınıfı varsa, bu sınıflar `cva` (Class Variance Authority) veya `clsx`/`tailwind-merge` gibi araçlarla organize edilmelidir.
+- **Component Isolation:** UI bileşenleri `apps/web/src/components/ui/` altında toplanmalı ve her bileşen kendi Tailwind sınıflarını barındırmalıdır.
+
+## 4. Performans ve Erişilebilirlik
+- **JIT Mode:** Her zaman Just-in-Time mode kullanılmalıdır.
+- **Contrast:** Tema renkleri seçilirken WCAG AA standartlarına uyulmalıdır.

package/templates/standards/testing-standards.md

@@ -0,0 +1,31 @@
+# 🧪 Kurumsal Test Standartları
+
+Bu belge, Agent Enderun tarafından yönetilen projelerdeki test disiplini ve kapsam kurallarını tanımlar. Kod kalitesi, otomatik testlerle güvence altına alınmalıdır.
+
+## 1. Test Piramidi ve Stratejisi
+- **Unit Tests (Birim Testler):** İş mantığının (Business Logic) en küçük parçaları için yazılır. Hızlı ve izole olmalıdır.
+- **Integration Tests (Entegrasyon Testleri):** Servislerin veritabanı veya dış API'ler ile uyumunu denetler.
+- **E2E Tests (Uçtan Uca Testler):** Kritik kullanıcı akışlarını (Login, Checkout vb.) gerçek tarayıcı ortamında simüle eder.
+
+## 2. Yazım Kuralları ve İsimlendirme
+- **Framework:** Vitest (Birim/Entegrasyon) ve Playwright (E2E) standarttır.
+- **Dosya İsimlendirme:** `[modul-ismi].test.ts` formatı kullanılır.
+- **Pattern (Given-When-Then):**
+  ```typescript
+  it("should create a user when valid data is provided", async () => {
+    // Given (Hazırlık)
+    const userData = { ... };
+    // When (Eylem)
+    const result = await userService.create(userData);
+    // Then (Doğrulama)
+    expect(result.id).toBeDefined();
+  });
+  ```
+
+## 3. Kapsam ve Başarı Kriterleri
+- **Kritik Yol Kapsamı:** Auth, Ödeme ve Veri Mutasyonu süreçleri %100 test kapsamına alınmalıdır.
+- **Zero Mock Policy (Internal):** Proje içi servisler arasında mock yerine gerçek kontratlar ve test veritabanları kullanılmalıdır. Dış servisler (Stripe, Twilio vb.) mocklanabilir.
+
+## 4. Denetim
+- `@quality` ajanı, her yeni fonksiyonda ilgili test dosyasının güncellendiğini doğrular.
+- Testleri geçmeyen kodlar (`npm run test`) asla onaylanmaz.

package/src/modules/adapters/antigravity-cli.ts

@@ -1,25 +0,0 @@
-import path from "path";
-import { writeJsonFile } from "../../shared/fs.js";
-import type { AdapterConfig } from "./types.js";
-import { registerGlobalAntigravityPlugins } from "./shared.js";
-
-export const config: AdapterConfig = {
-    id: "antigravity-cli",
-    frameworkDir: ".agents",
-    shimFile: "AGENTS.md",
-    shimTemplate: "src/cli/shims/antigravity-cli.md",
-    role: "general",
-    templateDir: ".enderun",
-    nestedDirs: ["agents", "rules", "knowledge"],
-    rulesDir: ".agents/rules",
-    rulesExt: ".md"
-};
-
-export function postInit(projectRoot: string, mcpBlock: unknown): void {
-    writeJsonFile(path.join(projectRoot, config.frameworkDir, "mcp_config.json"), mcpBlock);
-    console.warn(`✅ Antigravity CLI MCP → ${config.frameworkDir}/mcp_config.json`);
-
-    // Global registration
-    registerGlobalAntigravityPlugins(mcpBlock);
-}
-

package/src/modules/adapters/claude.ts

@@ -1,36 +0,0 @@
-import path from "path";
-import { writeJsonFile } from "../../shared/fs.js";
-import { findClaudeConfigPath, addMcpServerToClaude } from "../../cli/utils/claude.js";
-import type { AdapterConfig } from "./types.js";
-
-export const config: AdapterConfig = {
-    id: "claude",
-    frameworkDir: ".claude",
-    shimFile: "CLAUDE.md",
-    shimTemplate: "src/cli/shims/claude.md",
-    role: "architect",
-    templateDir: ".enderun",
-    nestedDirs: ["agents", "rules", "knowledge"],
-    rulesDir: ".claude/rules",
-    rulesExt: ".md"
-};
-
-export function postInit(projectRoot: string, mcpBlock: unknown): void {
-    const configPath = findClaudeConfigPath();
-    if (configPath) {
-        const block = mcpBlock as { mcpServers: Record<string, unknown> };
-        // Build mcpEntry matching what adapters.ts does
-        const mcpEntry = block.mcpServers["agent-enderun"] as Record<string, unknown>;
-        const ok = addMcpServerToClaude(configPath, "agent-enderun", mcpEntry);
-        if (ok) {
-            console.warn(`✅ Claude MCP registered → ${configPath}`);
-        } else {
-            console.warn("⚠️  Could not patch Claude config automatically.");
-        }
-    } else {
-        console.warn("⚠️  Claude config not found. Add MCP manually (see README).");
-    }
-    writeJsonFile(path.join(projectRoot, config.frameworkDir, "mcp_config.json"), mcpBlock);
-    writeJsonFile(path.join(projectRoot, ".mcp.json"), mcpBlock);
-    console.warn("✅ Claude Code Project MCP → .mcp.json");
-}

package/src/modules/adapters/codex.ts

@@ -1,22 +0,0 @@
-import path from "path";
-import { writeJsonFile } from "../../shared/fs.js";
-import type { AdapterConfig } from "./types.js";
-
-export const config: AdapterConfig = {
-    id: "codex",
-    frameworkDir: ".github",
-    shimFile: "copilot-instructions.md",
-    shimTemplate: "src/cli/shims/codex.md",
-    role: "implementer",
-    templateDir: ".enderun",
-    nestedDirs: ["agents", "rules", "knowledge", "instructions"],
-    rulesDir: ".github/instructions",
-    rulesExt: ".md"
-};
-
-export function postInit(projectRoot: string, mcpBlock: unknown): void {
-    writeJsonFile(path.join(projectRoot, config.frameworkDir, "mcp_config.json"), mcpBlock);
-    writeJsonFile(path.join(projectRoot, ".vscode/mcp.json"), mcpBlock);
-    writeJsonFile(path.join(projectRoot, ".mcp.json"), mcpBlock);
-    console.warn("✅ GitHub Copilot Project MCP → .vscode/mcp.json & .mcp.json");
-}

package/src/modules/adapters/cursor.ts

@@ -1,22 +0,0 @@
-import path from "path";
-import { writeJsonFile } from "../../shared/fs.js";
-import type { AdapterConfig } from "./types.js";
-
-export const config: AdapterConfig = {
-    id: "cursor",
-    frameworkDir: ".cursor",
-    shimFile: "CURSOR.md",
-    shimTemplate: "src/cli/shims/cursor.mdc",
-    role: "implementer",
-    templateDir: ".enderun",
-    nestedDirs: ["agents", "rules", "knowledge"],
-    rulesDir: ".cursor/rules",
-    rulesExt: ".mdc"
-};
-
-export function postInit(projectRoot: string, mcpBlock: unknown): void {
-    const frameworkDir = config.frameworkDir;
-    // Native Cursor project-level MCP config
-    writeJsonFile(path.join(projectRoot, frameworkDir, "mcp.json"), mcpBlock);
-    console.warn(`✅ Cursor IDE Project MCP → ${frameworkDir}/mcp.json`);
-}

package/src/modules/adapters/gemini.ts

@@ -1,27 +0,0 @@
-import path from "path";
-import { writeJsonFile } from "../../shared/fs.js";
-import type { AdapterConfig } from "./types.js";
-import { registerGlobalAntigravityPlugins } from "./shared.js";
-
-export const config: AdapterConfig = {
-    id: "gemini",
-    frameworkDir: ".gemini",
-    shimFile: "GEMINI.md",
-    shimTemplate: "src/cli/shims/gemini.md",
-    role: "commander",
-    templateDir: ".enderun",
-    nestedDirs: ["agents", "rules", "knowledge"],
-    rulesDir: ".gemini/rules",
-    rulesExt: ".md"
-};
-
-export function postInit(projectRoot: string, mcpBlock: unknown): void {
-    const frameworkDir = config.frameworkDir;
-    // Standard project-level MCP config
-    writeJsonFile(path.join(projectRoot, frameworkDir, "mcp.json"), mcpBlock);
-    console.warn(`✅ Gemini MCP registered → ${frameworkDir}/mcp.json`);
-
-    // Global registration
-    registerGlobalAntigravityPlugins(mcpBlock);
-}
-

package/src/modules/adapters/grok.ts

@@ -1,20 +0,0 @@
-import path from "path";
-import { writeJsonFile } from "../../shared/fs.js";
-import type { AdapterConfig } from "./types.js";
-
-export const config: AdapterConfig = {
-    id: "grok",
-    frameworkDir: ".grok",
-    shimFile: "GROK.md",
-    shimTemplate: "src/cli/shims/grok.md",
-    role: "researcher",
-    templateDir: ".enderun",
-    nestedDirs: ["agents", "rules", "knowledge"],
-    rulesDir: ".grok/rules",
-    rulesExt: ".md"
-};
-
-export function postInit(projectRoot: string, mcpBlock: unknown): void {
-    writeJsonFile(path.join(projectRoot, config.frameworkDir, "mcp_config.json"), mcpBlock);
-    console.warn(`✅ Grok MCP → ${config.frameworkDir}/mcp_config.json`);
-}

package/templates/architecture/agents-manifest.md

@@ -1,79 +0,0 @@
-# 🎖️ Agent Enderun — Agent Army Manifest
-
-This document serves as the complete manifest of the 13 specialist agents forming the **Enderun Army**. Each agent is mapped to a specific capability index, toolset, and specialized directive under the Enderun Discipline (Osmanlı Nizami).
-
----
-
-## 🏛️ Agent Army Overview
-
-| Agent Name | Role | Capability | Primary Tools | Tags |
-| :--- | :--- | :--- | :--- | :--- |
-| **`@manager`** | Orchestration | 10/10 | `orchestrate_loop`, `send_agent_message` | `core`, `orchestration` |
-| **`@architect`** | System Design | 9/10 | `write_file`, `replace_text` | `core`, `design` |
-| **`@backend`** | Backend Logic | 9/10 | `replace_text`, `run_shell_command` | `core`, `logic` |
-| **`@frontend`** | User Interface | 9/10 | `replace_text`, `view_file` | `core`, `ui` |
-| **`@quality`** | Quality Audit | 9/10 | `run_shell_command`, `view_file` | `core`, `audit` |
-| **`@database`** | Data Management | 9/10 | `replace_text`, `run_shell_command` | `core`, `data` |
-| **`@devops`** | Infrastructure | 9/10 | `run_shell_command`, `view_file` | `core`, `infra` |
-| **`@explorer`** | Reconnaissance | 8/10 | `grep_search`, `list_dir`, `glob` | `core`, `recon` |
-| **`@git`** | Version Control | 9/10 | `run_shell_command`, `view_file` | `core`, `logistics` |
-| **`@mobile`** | Mobile Apps | 9/10 | `replace_text`, `view_file` | `core`, `mobile` |
-| **`@native`** | System-Level | 9/10 | `run_shell_command`, `replace_text` | `core`, `native` |
-| **`@security`** | Security Audit | 10/10 | `grep_search`, `run_shell_command` | `core`, `security` |
-| **`@analyst`** | Requirements | 9/10 | `grep_search`, `list_dir` | `core`, `strategy` |
-
----
-
-## 🤖 Specialist Directives
-
-### 1. `@manager` (Orchestrator)
-- **Directive:** Strategic commander of the army. Coordinates executions, schedules traces, and delegates tasks to specialists.
-- **Mandate:** Enforces Enderun discipline, inherits and passes Trace IDs, and handles final compliance approvals.
-
-### 2. `@architect` (Lead Architect)
-- **Directive:** Outlines initial codebase design, defines schema contracts, and manages the architectural blueprint.
-- **Mandate:** Enforces strict modular boundaries and verifies structural sanity.
-
-### 3. `@backend` (Backend Specialist)
-- **Directive:** Builds REST/GraphQL API systems, develops server components, and manages server business rules.
-- **Mandate:** Ensures robust error handling, schema boundaries, and server logic.
-
-### 4. `@frontend` (Frontend Specialist)
-- **Directive:** Builds user interfaces, manages layouts and states using modern styling (Panda CSS, CSS variables).
-- **Mandate:** Enforces design systems, accessibility, and high performance rendering.
-
-### 5. `@quality` (Quality Specialist)
-- **Directive:** Validates unit coverage, performs static analyses, and ensures code rules compliance.
-- **Mandate:** Runs ESLint rules, checks console logs, and audits project safety.
-
-### 6. `@database` (Database Specialist)
-- **Directive:** Designs relational/NoSQL schemas, writes migrations, and optimizes SQL query metrics.
-- **Mandate:** Guarantees data transaction integrity and index configurations.
-
-### 7. `@devops` (Infrastructure Specialist)
-- **Directive:** Establishes CI/CD pipelines, configures Docker environments, and tracks cloud deployments.
-- **Mandate:** Ensures high availability, automation scripts, and safe release cycles.
-
-### 8. `@explorer` (Intel Explorer)
-- **Directive:** Gathers codebase intelligence, discovers file dependencies, and maps codebase symbols.
-- **Mandate:** Guides planning agents by providing context graphs and code dependencies.
-
-### 9. `@git` (Logistics Master)
-- **Directive:** Handles atomic commit flows, merges code, resolves conflicts, and manages release tags.
-- **Mandate:** Enforces descriptive commit tags mapped to active Trace IDs.
-
-### 10. `@mobile` (Mobile Specialist)
-- **Directive:** Implements mobile interfaces using cross-platform layouts (React Native / Expo).
-- **Mandate:** Optimizes mobile assets, native performance, and state stores.
-
-### 11. `@native` (Native Division)
-- **Directive:** Interfaces with native operating system hooks, constructs desktop platforms, and manages hardware bridges.
-- **Mandate:** Exercises deep security protocols during system interactions.
-
-### 12. `@security` (Security Specialist)
-- **Directive:** Audits repositories for credentials/secrets, evaluates package dependencies, and verifies authentication flows.
-- **Mandate:** Rejects insecure actions, validates cryptographic keys, and enforces access control checks.
-
-### 13. `@analyst` (Business Analyst)
-- **Directive:** Maps requirements to codebase implementations, writes business specification rules, and audits contracts.
-- **Mandate:** Validates feature alignment against system specifications.

package/templates/architecture/approval-flows.md

@@ -1,61 +0,0 @@
-# Enterprise Approval Flows & Governance Protocol (v0.8.5)
-
-This protocol defines the strict governance model for high-risk operations and cross-agent approval flows within the Agent Enderun framework. It ensures that no critical mutations or sensitive actions are executed by AI specialists without formal orchestration and explicit authorization.
-
----
-
-## 🔒 1. High-Risk Action Catalog
-
-Specialist agents (@backend, @frontend, @devops, etc.) are strictly prohibited from performing any of the following administrative operations autonomously. If requested, they must halt work, shift to `WAITING` status, and escalate to `@manager` via the **Hermes Protocol**.
-
-| Category | High-Risk Operations | Designated Controller |
-| :--- | :--- | :--- |
-| **Data Management** | Bulk delete/purge, data migration, PII export, schema modifications. | `@manager` |
-| **System Security** | Security credentials modification, CSP rule changes, private keys rotation. | `@security` & `@manager` |
-| **Infrastructure** | Production deployment triggers, container orchestration changes, ENV vars rotation. | `@devops` & `@manager` |
-| **Contract Stability**| Breaking API modifications, major type updates inside `src/types/`. | `@backend` & `@manager` |
-
----
-
-## 💬 2. Hermes approval Flow (`managerApproval`)
-
-When a specialist agent encounters a high-risk operation, it must execute the following message queue protocol.
-
-### Step 1: Lock the Hermes Inbox
-To prevent race conditions, the agent must check for `{{FRAMEWORK_DIR}}/messages/.lock`. If the lock is active, the agent retries (up to 3 times, waiting 500ms between attempts). If no lock exists, it creates `.lock` to claim the channel.
-
-### Step 2: Write Approval Request File
-The requesting agent writes a structured JSON file inside `{{FRAMEWORK_DIR}}/queue/pending/` or `{{FRAMEWORK_DIR}}/messages/` with the unique `Trace ID` (ULID) and a `managerApproval` request flag:
-
-```json
-{
-  "traceId": "01JM5S8A0F2B4C6D8E0G2H4J6K",
-  "from": "@backend",
-  "to": "@manager",
-  "category": "ALERT",
-  "priority": "HIGH",
-  "action": "RUN_DATABASE_MIGRATION",
-  "payload": {
-    "migrationFile": "20260527_add_tenant_rls_policy.sql",
-    "rationale": "Enables multi-tenant Row Level Security (RLS) on PostgreSQL."
-  },
-  "timestamp": "2026-05-27T20:45:00Z"
-}
-```
-
-### Step 3: Shift Agent State
-The requesting agent updates its state inside `.enderun/STATUS.md` to `WAITING` and appends `BLOCKED_BY_APPROVAL` to its action log.
-
-### Step 4: Human-in-the-Loop Signing
-In production workspaces, the `@manager` compiles the war-room report and requests the human architect (CTO) to sign off on the transaction. The human developer signs the transaction by executing:
-```bash
-npx agent-enderun approve 01JM5S8A0F2B4C6D8E0G2H4J6K
-```
-Or by writing a signed token to the message queue.
-
----
-
-## 🚦 3. Conflict Resolution & Rollback
-
-1.  **Approval Rejection:** If the transaction is rejected, `@manager` issues a `REJECTED` signal. The specialist agent rolls back any staged changes and returns to `IDLE` state.
-2.  **Stale Lock Recovery:** If a Hermes inbox `.lock` remains for more than 1.5 seconds, the active agent assumes a **stale lock** (LLM timeout), deletes the lock, and logs the incident to `@manager`.

package/templates/architecture/enterprise-architecture.md

@@ -1,69 +0,0 @@
-# 🎖️ Agent Enderun — Enterprise Architecture & Standards
-
-This document establishes the architecture patterns, design guidelines, and code structures of the **Enderun Core Framework** for enterprise deployments.
-
----
-
-## 🏛️ Project Directory Structure
-
-```text
-agent-enderun/
-├── bin/                       # Executable CLI entrypoints
-├── docs/                      # Technical documentation & design guides
-├── framework-mcp/             # Modular MCP server implementation (workspace)
-│   ├── src/                   # MCP server source files
-│   └── tests/                 # Unit tests for MCP server tools
-├── src/                       # CLI & Core framework logic
-│   ├── cli/                   # Command line interface commands & adapters
-│   │   ├── adapters/          # Target-specific IDE adapters (VS Code, Cursor, etc.)
-│   │   ├── commands/          # Subcommand handlers (init, status, check)
-│   │   └── utils/             # CLI-specific utilities
-│   ├── modules/               # Core framework modules
-│   │   ├── adapters/          # Target environment adapter profiles
-│   │   ├── agents/            # Orchestrator & Specialist Agent definitions
-│   │   └── skills/            # Capability templates for specialist actions
-│   └── shared/                # Core shared business components
-│       ├── config.ts          # Zod environment configuration
-│       ├── container.ts       # Dependency injection container
-│       ├── errors.ts          # Standard typed domain errors
-│       ├── fs.ts              # File system helper wrappers
-│       └── logger.ts          # Enterprise logging subsystem
-```
-
----
-
-## ⚙️ Core Design Principles
-
-### 1. Unified Domain Errors
-All errors thrown by modules must inherit from `EnderunBaseError` in [src/shared/errors.ts](file:///Users/ybekar/Desktop/Projeler/agent-enderun/src/shared/errors.ts).
-This ensures:
-- Unique error codes (`CONFIGURATION_ERROR`, `VALIDATION_ERROR`, etc.)
-- Metadata enrichment via the `details` field
-- Structured JSON serialization in production environments
-
-### 2. Zod-Validated Environments
-Configuration options must never be read directly from raw environment variables (`process.env`). Instead, they are parsed and frozen inside [src/shared/config.ts](file:///Users/ybekar/Desktop/Projeler/agent-enderun/src/shared/config.ts) at initialization. If a required value is missing, the application crashes immediately with a clean explanation.
-
-### 3. Service Container (Dependency Injection)
-To write testable and decoupled code, singletons and factories are registered and resolved using the `ServiceContainer` defined in [src/shared/container.ts](file:///Users/ybekar/Desktop/Projeler/agent-enderun/src/shared/container.ts).
-```typescript
-import { container } from "./shared/container.js";
-
-// Register
-container.register("Logger", logger);
-
-// Resolve
-const logger = container.resolve<EnterpriseLogger>("Logger");
-```
-
-### 4. Structured Observability
-Use `EnterpriseLogger` in [src/shared/logger.ts](file:///Users/ybekar/Desktop/Projeler/agent-enderun/src/shared/logger.ts) for logging:
-- In **development**, logs are colorized and formatted for human readability.
-- In **production**, logs are structured as JSON for log aggregators (e.g. Datadog, ELK).
-
----
-
-## 🛡️ Code & Type Guidelines
-- **Strict Typing:** Always keep `"strict": true` in `tsconfig.json`. Explicit return types are required on all exports.
-- **Asynchronous Flow:** Use async/await over raw promise chains. Handle failures with localized try/catch blocks and wrap in domain errors.
-- **Atomic Operations:** Always use transactional or surgical replacements when editing configuration/code files.

package/templates/architecture/standards/crud-governance.md

@@ -1,46 +0,0 @@
-# Enterprise CRUD & Admin Governance Protocol (v1.0.0)
-
-This protocol defines the strict operational boundaries, coding practices, and routing rules for High-Risk CRUD (Create, Read, Update, Delete) operations and Administrative actions within Agent Enderun projects.
-
----
-
-## 🔒 1. High-Risk Action Classification
-
-To maintain security, database consistency, and prevent unauthorized system changes, actions are split into **Standard Operations** and **High-Risk Operations**.
-
-### 🛑 High-Risk Administrative Operations (Requires @manager Sign-Off)
-The following actions must **never** be performed by specialist agents (`@backend`, `@frontend`, `@database`, etc.) autonomously. They must halt execution and redirect the task to `@manager` via the Hermes Message Broker.
-
-1. **User & Permission Management:** Modifying roles, assigning admin rights, creating api keys.
-2. **Bulk Destructive Operations:** Truncating tables, purging databases, executing mass-deletions without strict tenant-isolation clauses.
-3. **Database Schema Alterations:** Running database migrations, altering column constraints, deleting tables.
-4. **PII & Data Export:** Bulk exporting personal user info, system settings, or database backups.
-5. **System Config & CSP Policies:** Modifying environment variables, CORS origins, or content security policies.
-6. **Billing & Financial Actions:** Integrating payment gateways, editing subscription rules.
-
----
-
-## 🛡️ 2. Specialist Redirection Policy
-
-When a specialist agent receives a request containing any High-Risk Administrative Operation, it must immediately return the following standard refusal message and halt code output:
-
-> **[REFUSAL - Corporate CRUD Governance Standard]**
-> I am not authorized to perform administrative actions or bulk mutations. 
-> I have redirected this request to the Orchestrator (@manager) under Trace ID: {{TRACE_ID}}.
-> Please execute the manager approval sequence to proceed.
-
-### Action Flow:
-1. **Detect** high-risk keyword or context in the task.
-2. **Create** an approval request using the `send_agent_message` tool targeting `@manager` with category `ALERT` and priority `HIGH`.
-3. **Update** state in `STATUS.md` to `WAITING` with note `BLOCKED_BY_CRUD_GOVERNANCE`.
-
----
-
-## 🧬 3. Safe Coding Standard (Read/Write Rules)
-
-For permitted Standard CRUD operations, the following rules are strictly enforced:
-
-1. **Isolated Repository Pattern:** All database reads and writes must be encapsulated in repository/service classes. Writing database queries directly inside fastify controllers or nextjs actions is **strictly forbidden**.
-2. **No Raw SQL:** All database mutations must use typed query builders (e.g. `Kysely`). Direct string-concatenated SQL queries are **strictly forbidden** to prevent SQL injection.
-3. **Branded Types:** Entities must use branded types (e.g. `UserId`, `ProjectId`) instead of primitive types (`string`, `number`) to avoid type mixing.
-4. **Soft Deletes by Default:** All business-domain deletions must use a `deletedAt` timestamp instead of raw SQL `DELETE` mutations, unless explicitly designated by `@manager`.

package/templates/architecture/standards/data-fetching-patterns.md

@@ -1,13 +0,0 @@
-# Architecture Patterns & Data Fetching
-
-## 1. UI Layout & Component Structure
-- **Shared UI:** All atomic components (Button, Input, etc.) must be created in `apps/web/src/components/ui/` using **Panda CSS**.
-- **Layouts:** Use wrapper layouts (`Shell`, `Workspace`) in `apps/web/src/components/layout/`. Every page must utilize these layouts to maintain consistency.
-- **Procedural Continuity:** Never re-create an existing component. Search the `components/ui/` directory first.
-
-## 2. Data Fetching & API (CRUD) Patterns
-- **Repository Pattern:** Backend logic must be separated into service/repository layers. No direct DB calls in controllers.
-- **Contract-First:** All API endpoints must use types imported from `apps/backend/src/types/`.
-- **API Layer:** Frontend must use typed fetcher functions located in `apps/web/src/api/`.
-- **Validation:** Use `zod` for request validation on the backend and response validation on the frontend.
-- **Audit Logging:** Every mutation (POST/PUT/PATCH/DELETE) MUST trigger an audit log service call with the current Trace ID.

package/templates/architecture/standards/design-system.md

@@ -1,31 +0,0 @@
-# Design System — Agent Enderun (Enterprise)
-
-## 1. Design Philosophy
-The Agent Enderun Design System focuses on **"Enterprise Clarity"**. Our interfaces are professional, minimalist tools designed to maximize data comprehension and operational speed. 
-
-- **Clarity:** Every pixel must serve a purpose. If it doesn't aid understanding or action, remove it.
-- **Consistency:** Uniform visual language across all applications.
-- **Responsiveness:** Fluid adaptation to any screen size, from mobile to ultra-wide monitors.
-
-## 2. Technical Requirements
-- **Styling:** Panda CSS is MANDATORY. Tailwind, CSS-in-JS, and raw CSS modules are FORBIDDEN.
-- **Zero UI Library:** No external components (MUI, AntD, Shadcn). All UI components must be built from scratch in `apps/web/src/components/ui/`.
-- **Tokenization:** All values (colors, spacing, typography) MUST use `panda.config.ts` tokens. NO hardcoded pixels.
-- **Responsiveness:** Mobile-First design is MANDATORY.
-
-## 3. Responsive Mandate (Mobile-First)
-- **Design Strategy:** Design for 320px width first.
-- **Breakpoints:**
-  - `sm`: 640px
-  - `md`: 768px
-  - `lg`: 1024px
-  - `xl`: 1280px
-  - `2xl`: 1536px
-- **Layouts:** Use Flexbox and Grid. Fixed widths are FORBIDDEN. Use `clamp()` for fluid typography and spacing where applicable.
-- **Reflow:** Content MUST gracefully wrap, stack, or collapse based on screen width without horizontal scrollbars.
-
-## 4. UI/UX Interaction Standards
-- **Interactions:** Hover, Active, and Disabled states are MANDATORY for all interactive elements.
-- **Feedback:** Async operations MUST show Loading states and Toast notifications on completion/error.
-- **Blocking Dialogues:** Native `alert()` and `confirm()` functions are **FORBIDDEN**. Use project-integrated `Modal` components for blocking user interactions and `Toaster` for non-blocking feedback.
-- **Focus:** All interactive elements must have a visible `focus` state for accessibility.

package/templates/architecture/standards/documentation-ownership.md

@@ -1,21 +0,0 @@
-# 🎖️ Agent Enderun — Documentation Ownership & Memory Protocol
-
-This document establishes the binding discipline for where project documentation is stored and how AI agents interact with it.
-
-## 🏛️ Core Principles
-
-1. **Human-AI Separation of Concerns:**
-   - **`docs/` (Root Directory):** Single Source of Truth (SSOT) for human intentions, business requirements, API specifications, and architectural standards.
-   - **`.gemini/` or `.agents/` (Framework Directory):** Operational workspace and memory of the AI agents. Contains state machines, task tracking, rules, and internal logs.
-
-2. **The Documentation Discipline:**
-   - The user clarifies requirements in the chat.
-   - Once requirements are finalized, they are documented in the root `docs/` directory (e.g., `docs/project-docs.md`, `docs/api/`, etc.) by the user or the agent.
-   - When building features or validating code, the agent **reads** the documents from `docs/` and uses them as the blueprint.
-   - The agent **must not** duplicate requirements inside the internal memory (`PROJECT_MEMORY.md`). The internal memory only tracks *what has been done* and *what is currently running* (Trace ID, active phase, history of implemented steps referencing the doc files).
-
-## 🛡️ Agent Rules
-
-- **Do Not Pollute Internal Memory:** Never copy full requirement specifications or code files into `.gemini/memory/PROJECT_MEMORY.md`. Refer to the root `docs/` files instead.
-- **Exclusivity:** All project-specific architectural records, API catalogs, database schemas, and user manuals must be created in the `docs/` folder.
-- **Read Before Code:** Before writing any application code or generating files, the agent must check the root `docs/` directory for context and specifications.

package/templates/architecture/standards/logging.md

@@ -1,7 +0,0 @@
-# Logging & Audit Standard
-
-- **No Console.log:** Production code MUST NOT use `console.log`, `console.warn`, or `console.error`. These are FORBIDDEN in the codebase.
-- **Structured Logging:** All logging MUST go through the centralized `Logger` service (`apps/backend/src/services/logger.ts`). Logs must be produced in JSON format.
-- **Traceability:** Every log entry MUST include the active `traceId` context to ensure cross-agent and cross-service traceability.
-- **Audit Logging:** Every state-mutating operation (POST, PUT, PATCH, DELETE) MUST trigger an `AuditService.log()` call, linking the action to the responsible agent, Trace ID, and entity ID.
-- **Log Persistence:** Logs must be persisted in an structured manner suitable for downstream ingestion by log management systems.