Zero to Production Orchestrator (2026 Master Edition)

English | Bahasa Indonesia

---

<a name="english"></a>

English

Orchestration & Integration

Connects and orchestrates with relevant domain skills like brainstorming, zero-to-prod-orchestrator, and session-memory-manager to ensure cohesive execution.

Overview

The Zero to Production Orchestrator is the ultimate master skill designed to orchestrate the entire vibes-plug ecosystem as a highly interconnected engineering swarm. By acting as the central conductor, it ensures that no sub-skill is executed in isolation. It guides fullstack developers through the complete software engineering lifecycle — from concept discovery to AI integration, multi-platform backend architecture, design systems, automated testing, GEO/AEO optimization, and production deployment.

Trigger Conditions

• Starting any new application development from scratch.
• Asking for a complete, end-to-end fullstack development roadmap.
• Orchestrating multiple domain skills across UI, Backend, AI/LLM, Database, Security, and Cloud.

Core Principles & Hard Gates

1. Never Skip Phases: Each phase must be completed and validated before advancing to the next.
2. Continuous Auto-Documentation: Invoke prd-architect to log changes in CHANGELOG.md and BLUEPRINT.md after every major milestone.
3. Strict Progress Tracking: Maintain a PROGRESS.md checklist in the repository root.
4. State Preservation & Context: Utilize session-memory-manager when pausing work or starting a new session to preserve full context.
5. Sovereign Anti-Slop Mandate: Zero-tolerance for placeholders (// TODO, // ...), mock data in prod, syntax comments, and sycophancy across all phases. Run node scripts/check-anti-slop.js --strict as a mandatory validation gate before phase completion.

---

🐝 Swarm Execution Topologies & Subagent Delegation Protocols (2026 Master)

Whenever executing complex multi-domain phases, the Orchestrator MUST act as Swarm Director and delegate to specialized subagents in parallel:

1. Topology Selection

• Fan-Out / Fan-In: Used in Phase 4 (APIs + MCP + DB) and Phase 5 (Frontend UI + Mobile + Animations).
  • Director dispatches Subagent A (Frontend UI via senior-frontend), Subagent B (Backend API via js-backend-expert), Subagent C (DB Migration via database-orm-expert).
  • Collects results and runs integration verification.
• Pipeline Saga: Sequential execution across Phases 1 to 8. Each phase produces a checkpoint artifact (CONTEXT_MAP.md, BLUEPRINT.md).
• Critic-Validator Gate: Used in Phase 6. Implementer agent submits code -> Auditor agent (coderabbit, autonomous-red-teamer) reviews -> Fixer agent (autonomous-tdd-debugger) resolves failures.

2. Subagent Context Passing Protocol

When delegating to any subagent, ALWAYS pass:

1. Exact scope and non-goals.
2. File paths of relevant schemas (schema.prisma / schema.ts) and API routes.
3. Path to PRD.md and BLUEPRINT.md.
4. Verification command (e.g. npm test, tsc --noEmit) to confirm completion before returning.

3. Sovereign Token Frugality & Context Pruning Protocol

To maintain token efficiency without sacrificing skill capabilities:

1. Selective Context Slicing: Never pass the entire global chat transcript to subagents. Pass only targeted interface types, local schema slices, and the active task boundary (CONTEXT_MAP.md slice).
2. Deterministic KV-Cache Prefix Locking: Ensure static prompt instructions, tool definitions, and skill contracts remain byte-identical across calls (kv-cache-prefix-optimizer) to maximize prompt caching hit rates (75-90% discount).
3. Adaptive Model Cascading (adaptive-model-cascade):
   • flash_lite / flash: Discovery, AST scans, file searches, lint checks, and localized test validations.
   • pro / inherit: Architectural synthesis, multi-domain schemas, and security red-teaming.
4. Ultra-Compact Phase Checkpointing (session-memory-manager):
   • At the completion of each phase, persist a compressed checkpoint (<300 tokens) in PROGRESS.md and CHECKPOINT.md.
5. Non-Degradation Invariant: Never truncate code with // TODO, // ..., or partial stubs to save tokens. Code quality and completeness must remain 100%.

---

The 8-Phase Master Fullstack Pipeline

  PHASE 1          PHASE 2          PHASE 3          PHASE 4
Discovery/PRD ---> Foundation ---> Database/ORM ---> Backend/APIs
      |                                                   |
      v                                                   v
  PHASE 8          PHASE 7          PHASE 6          PHASE 5
Launch/Deploy <--- Security/GEO <--- Testing/QA  <--- Frontend/UI

PHASE 1: Discovery & AI PRD Architectural Planning

Orchestrates: prd-architect, brainstorming, deep-research-analyst, mcp-server-architect, session-memory-manager, dependency-upgrade-migrator, app-analyzer-optimizer, seo, saas-architect, web-scraper, website-design-cloner, screenshot-to-code-expert, multimodal-spatial-video-cloner, headless-cms-expert, wordpress-headless-expert, documentation-site-expert, anti-slop, speculative-multi-draft-synthesizer, thinking-budget-orchestrator

• Conduct structured dialogue to clarify product intent, target audience, and non-functional goals.
• Automatically draft a comprehensive Product Requirements Document (PRD.md), Entity Relationship Diagram (ERD.md), and Documentation (DOCUMENTATION.md) alongside the Roadmap (ROADMAP.md).
• Determine dynamic thinking budget profile (thinking-budget-orchestrator) across speed (1,024t), balanced (4,096t), deep (32,768t), or exhaustive (65,536t).
• Plan AI/LLM integration strategy (Vercel AI SDK, MCP Server tools, or Multi-Agent Graph).
• For existing projects: audit dependency health and plan upgrades with dependency-upgrade-migrator.
• Initialize BLUEPRINT.md and PROGRESS.md.

PHASE 2: Project Foundation & Monorepo Setup

Orchestrates: monorepo-architect, micro-frontend-architect, bun-runtime-expert, python-programming-expert, go-programming-expert, ci-cd-devops-architect, spa-orchestrator, mpa-orchestrator, cloud-hosting-expert, mvc-expert, scalability-clean-code, api-design-expert, legacy-code-translator, web-3d-graphics-expert, web-game-engine-expert, glsl-shader-expert, webxr-ar-vr-expert, rust-programming-expert, typescript-expert, biome-linter-formatter-expert, composable-mach-architect, living-codebase-ast-graph, formal-spec-z3-verifier, structured-output-contracts

• Initialize monorepo (Turborepo + pnpm workspaces) or single repo foundation.
• Set up language runtimes: Node.js 24 LTS / Bun 1.2+ / Python 3.14+ (uv) / Go 1.25+ / Rust 2024.
• Configure Biome v2 (Rust-based linter + formatter) or Ruff, ESLint, Prettier, and TypeScript strict configurations.
• Establish strict JSON Schema and type contracts with structured-output-contracts.
• Setup initial CI/CD pipeline template (GitHub Actions, Docker).

PHASE 3: Database & Multi-Tenant Core Architecture

Orchestrates: fullstack-expert, database-orm-expert, domain-driven-design-expert, saas-multi-tenant, supabase-security-expert, data-pipeline-etl-expert, search-engine-expert, xerj-search-expert, geospatial-maps-expert, graph-rag-knowledge-expert

• Design normalized relational schemas, document models, and geospatial PostGIS structures.
• Configure ORM layer (Drizzle ORM / Prisma 6 / SQLx / sqlc) and full-text search indexing (Typesense/Meilisearch).
• Implement Row-Level Security (RLS) policies and tenant isolation.
• Apply initial database migrations and connection poolers (PgBouncer/Supavisor/Neon).

PHASE 4: Backend APIs, Microservices & AI Agents

Orchestrates: js-backend-expert, cloudflare-workers-expert, go-programming-expert, pydantic-ai-expert, vercel-ai-sdk-expert, frontier-ai-models-expert, synthetic-data-finetuning-expert, graphql-apollo-expert, openapi-swagger-codegen-expert, spec-driven-development-expert, ai-llm-integration-expert, ai-prompt-engineering-expert, ai-media-generation-expert, multi-agent-orchestration, mcp-server-architect, authentication-identity-expert, email-notification-expert, cron-scheduler-expert, rate-limit-abuse-prevention, file-upload-media-expert, saas-billing, agentic-micro-economy-architect, agentic-m2m-settlement, payment-gateway-expert, vector-db-rag-expert, async-queue-temporal-expert, doku-mcp-server, xerj-search-expert, event-driven-architect, gemini-agent-booster, realtime-collaboration-expert, api-gateway-proxy-expert, wasm-edge-computing-expert, sse-websocket-streaming-expert, n8n-automation-expert, chatbot-messaging-expert, pdf-document-generation-expert, ecommerce-expert, blockchain-web3-expert, local-slm-edge-ai-expert, voice-ai-realtime-agent, affective-computing-emotion-ai, graph-rag-knowledge-expert, browser-automation-expert, agentic-memory-architect, llm-finops-router, test-time-compute-optimizer, kv-cache-prefix-optimizer, speculative-multi-draft-synthesizer, autonomous-api-drift-healer, ephemeral-wasm-sandbox-executor, context-window-engineer, adaptive-model-cascade, llm-observability-expert, prompt-injection-firewall, thinking-budget-orchestrator, structured-output-contracts, parallel-tool-orchestrator

• Build high-throughput REST / GraphQL / gRPC APIs using Fastify 5, NestJS, Hono, Gin, or Axum.
• Implement authentication (Clerk, Auth.js, Supabase Auth) and RBAC middleware.
• Build MCP Server tools or stateful LangGraph multi-agent workflows with human-in-the-loop gates.
• Orchestrate concurrent parallel tool execution (parallel-tool-orchestrator) with vibes_batch and Promise.allSettled.
• Enforce strict schema guarantees and zero-drift output types (structured-output-contracts).
• Allocate dynamic thinking token budgets (thinking-budget-orchestrator) per subagent complexity.
• Integrate AI media generation (Flux/ElevenLabs/Whisper), chatbot platforms (WhatsApp/Telegram/Discord), and n8n automations.
• Implement context window engineering (sliding window, retrieval-augmented injection) and dynamic model cascading (adaptive-model-cascade, context-window-engineer).
• Integrate prompt injection defenses, input sanitization, and output guardrails (prompt-injection-firewall).
• Set up LLM observability, latency tracking, hallucination monitoring, and token spend telemetry (llm-observability-expert).
• Implement background processing queues (BullMQ + Redis), scheduled jobs, and rate limiters.
• Set up transactional email pipeline (Resend/Postmark) and PDF generation.
• Implement file upload with presigned URLs (S3/R2) and media processing.
• Deploy Computer-Using Agent (CUA) patterns for autonomous browser interactions.
• Setup continuous multimodal streaming API for realtime voice and vision.
• Orchestrate narrative simulation agents with human-like behavioral modeling.
• Integrate episodic memory system for long-term agent context retention.

PHASE 5: Frontend, Design Systems & Mobile Apps

Orchestrates: modern-web-guidance, design-system-architect, senior-frontend, anti-slop, vercel-ai-sdk-expert, nextjs-app-router-expert, vue-frontend-expert, astro-framework-expert, svelte-sveltekit-expert, solidjs-expert, angular-expert, tailwind-expert, tanstack-query-expert, spa-orchestrator, mobile-expo-expert, flutter-dart-expert, htmx-hypermedia-expert, apple-ecosystem-expert, tauri-expert, desktop-electron-expert, browser-extension-mv3-pro, form-validation-expert, svg-animation-motion-expert, app-promo-media-expert, web-3d-graphics-expert, web-game-engine-expert, glsl-shader-expert, webxr-ar-vr-expert, browser-automation-expert, hig, global-a11y-i18n-expert, state-management-expert, ui-ux-pro-max, affective-computing-emotion-ai, data-visualization-expert, rich-text-editor-expert, documentation-site-expert, blockchain-web3-expert, modern-css-native-expert, pwa-offline-first-expert, ephemeral-generative-ui-architect, multimodal-spatial-video-cloner, screenshot-to-code-expert, visual-ui-auditor, parallel-tool-orchestrator

• MANDATORY: Run modern-web-guidance FIRST before implementing any frontend HTML/CSS/JS features to ensure compliance with modern standards.
• Enforce Sovereign Anti-Slop UI standards (anti-slop): eradicate generic gradients, bento defaults, excessive glassmorphism, decorative dots/badges, and fake stat cards.
• Audit rendered UI elements against Pillar 7 standards with multimodal vision critic (visual-ui-auditor).
• Implement design tokens (OKLCH) and Tailwind CSS v4 @theme directive tokens.
• Construct accessible component primitives using Radix UI / Base UI and CVA variants.
• Convert UI screenshots, Figma frames, or whiteboard sketches into pixel-perfect frontend components with screenshot-to-code-expert.
• Integrate data visualizations (Recharts/Tremor/D3) and rich text editors (Tiptap/Lexical).
• Build React 19 / Next.js 15, Vue 3, Astro 5, Svelte 5, SolidJS 2, or Angular 19+ apps with state management and Web3 wallets.
• MANDATORY: Automatically scaffold standard pages: About, Profile, Contact, Terms of Reference/Service, and Privacy Policy.
• Implement complex forms with React Hook Form + Zod validation.
• If 3D Web or Web Games: architect with WebGPU (WebGPURenderer, PlayCanvas, or Babylon.js), KTX2 Basis Universal texture compression, Meshopt geometry, fixed-timestep physics loops, zero-GC object pooling, and 3D spatial audio.
• If SPA architecture — coordinate with spa-orchestrator for routing (TanStack Router), state (TanStack Query v5), and decoupled API layer.
• Integrate frontend state management with TanStack Query v5.

PHASE 6: Automated Testing, Error Resilience & Security Audit

Orchestrates: e2e-testing-expert, accessibility-testing-expert, autonomous-red-teamer, firebase-security-expert, error-resilience-expert, logging-error-tracking-expert, anti-slop, coderabbit, autonomous-tdd-debugger, browser-automation-expert, zero-trust-secret-vault, post-quantum-crypto-migrator, compliance-gdpr-privacy-expert, ai-safety-governance-expert, agentic-coding-workflow-expert, zero-tech-debt-auditor, living-codebase-ast-graph, ephemeral-wasm-sandbox-executor, formal-spec-z3-verifier, ai-code-review-autonomous, prompt-injection-firewall, property-mutation-testing-expert, spec-driven-development-expert, mcp-sampling-critic, visual-ui-auditor

• Write unit and integration tests with Vitest and pytest.
• Run autonomous multi-pass AI code self-review (ai-code-review-autonomous) across syntax, logic correctness, and security.
• Execute sampling/createMessage Critic-in-the-Loop review gates (mcp-sampling-critic) before merging or deployment.
• Execute visual UI regression audits (visual-ui-auditor) against Pillar 7 design standards.
• Execute property-based testing and mutation testing (property-mutation-testing-expert) with fast-check and Stryker to verify test suite invariants.
• Execute security fuzz testing (Atheris / cargo-fuzz / native Go fuzzing).
• Implement Error Boundaries, retry patterns, circuit breakers, and graceful degradation.
• Set up structured logging (Pino) and error tracking (Sentry) with source map uploads.
• Audit CORS, CSP headers, rate-limiting, and input sanitization.
• Run automated Anti-Slop Audit (node scripts/check-anti-slop.js --strict) to eliminate placeholders, syntax comments, and mock data before user handoff.
• Run Zero Technical Debt audit (zero-tech-debt-auditor) for dead code elimination (Knip protocol), DRY enforcement, and TypeScript strictness.

PHASE 7: DevOps, Deployment & Proactive Monitoring

Orchestrates: ci-cd-devops-architect, github-actions-expert, docker-container-expert, cloud-hosting-expert, performance-web-vitals, logging-error-tracking-expert, production-ready-hardener, proactive-background-watcher, data-telemetry-expert, feature-flag-analytics-expert, error-resilience-expert, self-healing-cloud-orchestrator, llm-observability-expert

• Perform pre-launch audit across Core Web Vitals (LCP, INP, CLS) and bundle sizes.
• Configure production monitoring, alerting rules, and on-call notifications.
• Configure LLM observability dashboards (Langfuse/Helicone) and production token spend alerting (llm-observability-expert).
• Optimize Generative Engine Optimization (GEO) for AI Overviews, Perplexity, ChatGPT Search, and deploy /llms.txt.
• Generate structured Schema.org JSON-LD markup and AEO conversion landing pages.

PHASE 8: Launch, Deployment & Handover

Orchestrates: cloud-hosting-expert, saas-billing, agentic-micro-economy-architect, agentic-m2m-settlement, saas-architect, prd-architect, ci-cd-devops-architect, doku-payment-gateway, payment-gateway-expert, app-promo-media-expert, zero-tech-debt-auditor, autonomous-api-drift-healer

• Deploy backend and edge services to Vercel, Cloudflare, AWS, or Railway.
• For SaaS applications: deploy Super Admin dashboard on a separate subdomain (e.g., admin.yourdomain.com) with strict role-based access (isSuperAdmin flag).
• Configure Stripe / Polar.sh / LemonSqueezy / DOKU SNAP BI billing and webhooks.
• Generate promotional media kit: App Store/Play Store screenshots, dynamic OpenGraph banners, and launch promo video teasers with app-promo-media-expert.
• Verify zero technical debt clearance (zero-tech-debt-auditor) before final handover.
• Finalize CHANGELOG.md, BLUEPRINT.md, and PROGRESS.md.
• Handover the production-grade application to the user.

---

<a name="bahasa-indonesia"></a>

Bahasa Indonesia

Integrasi Orkestrasi

Terhubung dan mengorkestrasi skill domain yang relevan seperti brainstorming, zero-to-prod-orchestrator, dan session-memory-manager untuk memastikan eksekusi yang kohesif.

Ringkasan

Zero to Production Orchestrator adalah skill master utama yang dirancang untuk mengorkestrasi seluruh ekosistem vibes-plug sebagai sebuah engineering swarm yang saling terhubung erat. Dengan bertindak sebagai konduktor pusat, skill ini memastikan tidak ada sub-skill yang dieksekusi secara terisolasi. Skill ini memandu pengembang fullstack melalui seluruh siklus hidup rekayasa perangkat lunak — mulai dari tahap ide awal hingga integrasi AI, arsitektur backend multi-platform, design system, pengujian otomatis, optimasi GEO/AEO, dan deployment produksi.

Kondisi Pemicu

• Memulai pengembangan aplikasi baru dari nol.
• Meminta panduan roadmap pengembangan fullstack end-to-end yang terstruktur.
• Mengorkestrasi berbagai skill spesialis lintas domain (UI, Backend, AI/LLM, Database, Keamanan, dan Cloud).

Prinsip Inti & Gerbang Ketat

1. Jangan Pernah Melewati Fase: Setiap fase harus diselesaikan dan divalidasi sebelum beralih ke fase berikutnya.
2. Otomatisasi Dokumentasi: Panggil prd-architect untuk memperbarui CHANGELOG.md dan BLUEPRINT.md setelah setiap milestone utama.
3. Pelacakan Progres: Pelihara daftar periksa PROGRESS.md di root repositori.
4. Preservasi State & Konteks: Gunakan session-memory-manager saat menjeda pekerjaan atau memulai sesi baru untuk menjaga konteks penuh.
5. Amanat Anti-Slop Berdaulat: Nol toleransi terhadap placeholder (// TODO, // ...), data mock di produksi, komentar sintaksis, dan basa-basi robotik di seluruh fase. Jalankan node scripts/check-anti-slop.js --strict sebagai gate verifikasi wajib sebelum penyelesaian fase.

---

🐝 Topologi Eksekusi Swarm & Protokol Delegasi Subagent (Master 2026)

Saat mengeksekusi fase multi-domain yang kompleks, Orkestrator WAJIB bertindak sebagai Swarm Director dan mendelegasikan tugas ke subagent spesialis secara paralel:

1. Pemilihan Topologi

• Fan-Out / Fan-In: Digunakan pada Fase 4 (API + MCP + DB) dan Fase 5 (Frontend UI + Mobile + Animasi).
  • Director mengirimkan tugas ke Subagent A (Frontend UI via senior-frontend), Subagent B (Backend API via js-backend-expert), Subagent C (Migrasi DB via database-orm-expert).
  • Mengumpulkan hasil dan menjalankan verifikasi integrasi.
• Pipeline Saga: Eksekusi berurutan dari Fase 1 hingga 8. Setiap fase menghasilkan artefak checkpoint (CONTEXT_MAP.md, BLUEPRINT.md).
• Critic-Validator Gate: Digunakan pada Fase 6. Agen implementor menyerahkan kode -> Agen auditor (coderabbit, autonomous-red-teamer) mereview -> Agen perbaikan (autonomous-tdd-debugger) menuntaskan kegagalan pengujian secara mandiri.

2. Protokol Pengiriman Konteks Subagent

Saat mendelegasikan tugas ke subagent, SELALU berikan:

1. Ruang lingkup tugas dan non-goals yang tegas.
2. Path file skema terkait (schema.prisma / schema.ts) dan route API.
3. Lokasi dokumen panduan PRD.md dan BLUEPRINT.md.
4. Perintah verifikasi (seperti npm test, tsc --noEmit) untuk memastikan kode tervalidasi sebelum kembali ke Director.

3. Protokol Efisiensi Token & Pemangkasan Konteks (Hemat Token)

Untuk menjaga efisiensi token tanpa mengurangi kapabilitas dan kedalaman fungsional skill:

1. Pemangkasan Konteks Selektif (Selective Slicing): Jangan pernah mengirimkan seluruh transkrip obrolan global ke subagent. Kirimkan hanya tipe interface terkait, irisan skema lokal, dan batasan tugas yang relevan (CONTEXT_MAP.md slice).
2. Penguncian Prefix KV-Cache Deterministik (kv-cache-prefix-optimizer): Pastikan instruksi statis, definisi tools, dan kontrak antarmuka tetap konsisten pada tingkat byte agar tingkat keberhasilan prompt cache mencapai 75-90%.
3. Kaskade Model Dinamis (adaptive-model-cascade):
   • flash_lite / flash: Pencarian file, pemindaian simbol AST, verifikasi linter, dan pengecekan sintaks.
   • pro / inherit: Desain arsitektur multi-domain, pembuktian invarian Z3, dan audit keamanan mendalam.
4. Checkpoint Fase Ultra-Ringkas (session-memory-manager):
   • Di akhir setiap fase, simpan checkpoint terkompresi (<300 token) di PROGRESS.md dan CHECKPOINT.md.
5. Hukum Mutlak Non-Degradasi: Dilarang memotong kode dengan // TODO, // ..., atau stub parsial demi menghemat token. Kualitas kode dan kelengkapan logika wajib 100%.

---

Master Pipeline Fullstack 8-Fase

FASE 1: Discovery & Perencanaan Arsitektur PRD AI

Mengorkestrasi: prd-architect, brainstorming, deep-research-analyst, mcp-server-architect, session-memory-manager, dependency-upgrade-migrator, app-analyzer-optimizer, seo, saas-architect, web-scraper, website-design-cloner, screenshot-to-code-expert, multimodal-spatial-video-cloner, headless-cms-expert, wordpress-headless-expert, documentation-site-expert, anti-slop, speculative-multi-draft-synthesizer, thinking-budget-orchestrator

• Dialog terstruktur untuk memperjelas tujuan produk, audiens target, dan persyaratan non-fungsional.
• Secara otomatis menyusun Product Requirements Document (PRD.md), Entity Relationship Diagram (ERD.md), dan Dokumentasi (DOCUMENTATION.md) yang komprehensif beserta Roadmap (ROADMAP.md).
• Tentukan profil thinking budget dinamis (thinking-budget-orchestrator) antara speed (1.024t), balanced (4.096t), deep (32.768t), atau exhaustive (65.536t).
• Merencanakan integrasi AI/LLM (Vercel AI SDK, alat MCP Server, atau Graf Multi-Agen).
• Untuk proyek yang sudah ada: audit kesehatan dependensi dan rencanakan upgrade dengan dependency-upgrade-migrator.
• Menginisialisasi BLUEPRINT.md dan PROGRESS.md.

FASE 2: Pondasi Proyek & Setup Monorepo

Mengorkestrasi: monorepo-architect, micro-frontend-architect, bun-runtime-expert, python-programming-expert, go-programming-expert, ci-cd-devops-architect, spa-orchestrator, mpa-orchestrator, cloud-hosting-expert, mvc-expert, scalability-clean-code, api-design-expert, legacy-code-translator, web-3d-graphics-expert, web-game-engine-expert, glsl-shader-expert, webxr-ar-vr-expert, rust-programming-expert, typescript-expert, biome-linter-formatter-expert, composable-mach-architect, living-codebase-ast-graph, formal-spec-z3-verifier, structured-output-contracts

• Inisialisasi monorepo (Turborepo + pnpm workspaces) atau repositori tunggal.
• Menyiapkan runtime bahasa: Node.js 24 LTS / Bun 1.2+ / Python 3.14+ (uv) / Go 1.25+ / Rust 2024.
• Konfigurasi Biome v2 (Rust linter & formatter) atau Ruff, ESLint, Prettier, dan TypeScript ketat.
• Tegakkan kontrak tipe skema JSON Schema terstruktur strict dengan structured-output-contracts.
• Menyiapkan template pipeline CI/CD awal (GitHub Actions, Docker).

FASE 3: Database & Arsitektur Multi-Tenant

Mengorkestrasi: fullstack-expert, database-orm-expert, domain-driven-design-expert, saas-multi-tenant, supabase-security-expert, data-pipeline-etl-expert, search-engine-expert, xerj-search-expert, geospatial-maps-expert, graph-rag-knowledge-expert

• Merancang skema relasional ter-normalisasi, pemodelan dokumen, dan struktur geospasial PostGIS.
• Konfigurasi lapisan ORM (Drizzle ORM / Prisma 6 / SQLx / sqlc) dan indexing mesin pencari (Typesense/Meilisearch).
• Mengimplementasikan kebijakan Row-Level Security (RLS) dan isolasi tenant.
• Menerapkan migrasi database awal dan connection poolers (PgBouncer/Supavisor/Neon).

FASE 4: Backend API, Microservices & Agen AI

Mengorkestrasi: js-backend-expert, cloudflare-workers-expert, go-programming-expert, pydantic-ai-expert, vercel-ai-sdk-expert, frontier-ai-models-expert, synthetic-data-finetuning-expert, graphql-apollo-expert, openapi-swagger-codegen-expert, spec-driven-development-expert, ai-llm-integration-expert, ai-prompt-engineering-expert, ai-media-generation-expert, multi-agent-orchestration, mcp-server-architect, authentication-identity-expert, email-notification-expert, cron-scheduler-expert, rate-limit-abuse-prevention, file-upload-media-expert, saas-billing, agentic-micro-economy-architect, agentic-m2m-settlement, payment-gateway-expert, vector-db-rag-expert, async-queue-temporal-expert, doku-mcp-server, xerj-search-expert, event-driven-architect, gemini-agent-booster, realtime-collaboration-expert, api-gateway-proxy-expert, wasm-edge-computing-expert, sse-websocket-streaming-expert, n8n-automation-expert, chatbot-messaging-expert, pdf-document-generation-expert, ecommerce-expert, blockchain-web3-expert, local-slm-edge-ai-expert, voice-ai-realtime-agent, affective-computing-emotion-ai, graph-rag-knowledge-expert, browser-automation-expert, agentic-memory-architect, llm-finops-router, test-time-compute-optimizer, kv-cache-prefix-optimizer, speculative-multi-draft-synthesizer, autonomous-api-drift-healer, ephemeral-wasm-sandbox-executor, context-window-engineer, adaptive-model-cascade, llm-observability-expert, prompt-injection-firewall, thinking-budget-orchestrator, structured-output-contracts, parallel-tool-orchestrator

• Bangun API REST / GraphQL / gRPC high-throughput menggunakan Fastify 5, NestJS, Hono, Gin, atau Axum.
• Mengimplementasikan autentikasi (Clerk, Auth.js, Supabase Auth) dan middleware RBAC.
• Membangun alat MCP Server atau alur kerja multi-agen LangGraph berbasis state dengan gerbang human-in-the-loop.
• Orkestrasikan eksekusi paralel multi-tool concurrent (parallel-tool-orchestrator) dengan vibes_batch dan Promise.allSettled.
• Tegakkan penjaminan skema terstruktur zero-drift (structured-output-contracts).
• Alokasikan budget token penalaran dinamis (thinking-budget-orchestrator) sesuai kompleksitas subagent.
• Mengintegrasikan generasi media AI (Flux/ElevenLabs/Whisper), bot perpesanan (WhatsApp/Telegram/Discord), dan otomasi n8n.
• Implementasikan rekayasa context window (sliding window, injeksi berbasis retrieval) dan kaskade model cerdas (context-window-engineer, adaptive-model-cascade).
• Integrasikan pertahanan prompt injection dan guardrail input/output runtime (prompt-injection-firewall).
• Siapkan observabilitas LLM, pelacakan latensi, pemantauan halusinasi, dan analitik biaya token (llm-observability-expert).
• Mengimplementasikan antrean pemrosesan latar belakang (BullMQ + Redis), tugas terjadwal, dan rate limiter.
• Menyiapkan pipeline email transaksional (Resend/Postmark), pembuatan PDF, dan sistem notifikasi.
• Mengimplementasikan upload file dengan presigned URL (S3/R2) dan pemrosesan media.
• Menerapkan pola Computer-Using Agent (CUA) untuk otomatisasi interaksi peramban (browser).
• Mengonfigurasi API streaming multimodal kontinu untuk suara dan visi real-time.
• Mengorkestrasi agen simulasi naratif dengan pemodelan perilaku mirip manusia.
• Mengintegrasikan sistem memori episodik untuk retensi konteks agen jangka panjang.

FASE 5: Frontend, Design System & Mobile App

Mengorkestrasi: modern-web-guidance, design-system-architect, senior-frontend, anti-slop, vercel-ai-sdk-expert, nextjs-app-router-expert, vue-frontend-expert, astro-framework-expert, svelte-sveltekit-expert, solidjs-expert, angular-expert, tailwind-expert, tanstack-query-expert, spa-orchestrator, mobile-expo-expert, flutter-dart-expert, htmx-hypermedia-expert, apple-ecosystem-expert, tauri-expert, desktop-electron-expert, browser-extension-mv3-pro, form-validation-expert, svg-animation-motion-expert, app-promo-media-expert, web-3d-graphics-expert, web-game-engine-expert, glsl-shader-expert, webxr-ar-vr-expert, browser-automation-expert, hig, global-a11y-i18n-expert, state-management-expert, ui-ux-pro-max, affective-computing-emotion-ai, data-visualization-expert, rich-text-editor-expert, documentation-site-expert, blockchain-web3-expert, modern-css-native-expert, pwa-offline-first-expert, ephemeral-generative-ui-architect, multimodal-spatial-video-cloner, screenshot-to-code-expert, visual-ui-auditor, parallel-tool-orchestrator

• MANDATORY: Jalankan modern-web-guidance PERTAMA KALI sebelum mengimplementasikan fitur frontend HTML/CSS/JS untuk memastikan kepatuhan dengan standar modern Google.
• Tegakkan standar Anti-Slop UI Berdaulat (anti-slop): basmi gradien klise, bento default, glassmorphism berlebih, titik status dekoratif, dan metrik tiruan.
• Audit elemen UI yang dirender terhadap standar Pillar 7 dengan penguji visi multimodal (visual-ui-auditor).
• Implementasikan token desain (OKLCH) dan konfigurasi tema Tailwind CSS v4.
• Bangun komponen primitif aksesibel menggunakan Radix UI / Base UI dan CVA.
• Konversi screenshot antarmuka, frame Figma, atau sketsa whiteboard menjadi komponen frontend siap produksi dengan screenshot-to-code-expert.
• Integrasikan visualisasi data (Recharts/Tremor/D3) dan editor rich text (Tiptap/Lexical).
• Buat halaman React 19 / Next.js 15, Vue 3, Astro 5, Svelte 5, SolidJS 2, atau Angular 19+ dengan wallet Web3.
• MANDATORY: Otomatis buat halaman standar: About, Profile, Contact, Terms of Reference/Service, dan Privacy Policy.
• Mengimplementasikan formulir kompleks dengan React Hook Form + validasi Zod.
• Jika Web 3D atau Web Game: arsitekturkan dengan WebGPU (WebGPURenderer, PlayCanvas, atau Babylon.js), kompresi tekstur KTX2 Basis Universal, geometri Meshopt, loop fisika fixed-timestep, zero-GC object pooling, dan audio spasial 3D.
• Jika arsitektur SPA — koordinasikan dengan spa-orchestrator untuk routing (TanStack Router), state (TanStack Query v5), dan API layer terpisah.
• Mengintegrasikan manajemen state frontend dengan TanStack Query v5.

FASE 6: Pengujian Otomatis, Ketahanan Error & Audit Keamanan

Mengorkestrasi: e2e-testing-expert, accessibility-testing-expert, autonomous-red-teamer, firebase-security-expert, error-resilience-expert, logging-error-tracking-expert, anti-slop, coderabbit, autonomous-tdd-debugger, browser-automation-expert, zero-trust-secret-vault, post-quantum-crypto-migrator, compliance-gdpr-privacy-expert, ai-safety-governance-expert, agentic-coding-workflow-expert, zero-tech-debt-auditor, living-codebase-ast-graph, ephemeral-wasm-sandbox-executor, formal-spec-z3-verifier, ai-code-review-autonomous, prompt-injection-firewall, property-mutation-testing-expert, spec-driven-development-expert, mcp-sampling-critic, visual-ui-auditor

• Menulis unit test dan integration test dengan Vitest dan pytest.
• Jalankan self-review kode otonom multi-pass (ai-code-review-autonomous) mencakup validasi sintaks, logika, arsitektur, dan keamanan.
• Eksekusi gerbang validasi review Critic-in-the-Loop berbasis sampling/createMessage (mcp-sampling-critic) sebelum merge atau rilis.
• Eksekusi audit regresi visual antarmuka (visual-ui-auditor) terhadap standar kepatuhan Pillar 7.
• Eksekusi property-based testing dan mutation testing (property-mutation-testing-expert) dengan fast-check dan Stryker untuk validasi invarian sistem.
• Jalankan pengujian prompt injection dan mitigasi serangan jailbreak dengan prompt-injection-firewall.
• Menulis pengujian browser E2E Playwright dan pengujian aksesibilitas WCAG 2.2 otomatis (@axe-core/playwright dan Pa11y).
• Menjalankan pengujian fuzzing keamanan (Atheris / cargo-fuzz / native Go fuzzing).
• Mengimplementasikan Error Boundary, pola retry, circuit breaker, dan degradasi anggun.
• Menyiapkan logging terstruktur (Pino) dan pelacakan error (Sentry) dengan upload source map.
• Mengaudit CORS, CSP headers, rate-limiting, dan sanitasi input.
• Menjalankan audit Anti-Slop otomatis (node scripts/check-anti-slop.js --strict) untuk membersihkan placeholder, komentar sintaksis, dan data tiruan sebelum serah terima ke pengguna.
• Menjalankan audit Zero Technical Debt (zero-tech-debt-auditor) untuk eliminasi kode mati (protokol Knip), penegakan DRY, dan ketelitian TypeScript.

FASE 7: Hardening Pra-Peluncuran, Monitoring & DevOps Sentinel

Mengorkestrasi: ci-cd-devops-architect, github-actions-expert, docker-container-expert, cloud-hosting-expert, performance-web-vitals, logging-error-tracking-expert, production-ready-hardener, proactive-background-watcher, data-telemetry-expert, feature-flag-analytics-expert, error-resilience-expert, self-healing-cloud-orchestrator, llm-observability-expert

• Audit pra-peluncuran pada Core Web Vitals (LCP, INP, CLS) dan ukuran bundle.
• Mengonfigurasi monitoring produksi, aturan alerting, dan notifikasi on-call.
• Konfigurasi dashboard observabilitas LLM (Langfuse/Helicone) dan pemantauan pengeluaran token produksi (llm-observability-expert).
• Mengoptimalkan GEO untuk AI Overviews, Perplexity, ChatGPT Search, dan merilis /llms.txt.
• Membuat markup terstruktur Schema.org JSON-LD dan landing page konversi AEO.

FASE 8: Peluncuran, Deployment & Serah Terima

Mengorkestrasi: cloud-hosting-expert, saas-billing, agentic-micro-economy-architect, agentic-m2m-settlement, saas-architect, prd-architect, ci-cd-devops-architect, doku-payment-gateway, payment-gateway-expert, app-promo-media-expert, zero-tech-debt-auditor

• Deploy backend dan edge services ke Vercel, Cloudflare, AWS, atau Railway.
• Untuk aplikasi SaaS: deploy dashboard Super Admin pada subdomain terpisah (misal: admin.domain.com) dengan kontrol akses berbasis role (isSuperAdmin).
• Konfigurasi billing Stripe / Polar.sh / LemonSqueezy / DOKU SNAP BI dan webhooks.
• Memproduksi paket media promosi: screenshot App Store/Google Play, banner Open Graph dinamis, dan video teaser peluncuran dengan app-promo-media-expert.
• Verifikasi kelulusan audit Zero Technical Debt (zero-tech-debt-auditor) sebelum rilis final.
• Menyelesaikan CHANGELOG.md, BLUEPRINT.md, dan PROGRESS.md.
• Serah terima aplikasi siap produksi kepada pengguna.

---

🎨 Automatic Visual Assets Generation Mandate (CRITICAL)

MANDATORY: Whenever you are building a new application, scaffolding a project, or finalizing the initial UI/UX, you MUST automatically use the generate_image tool to create a custom logo that perfectly matches the application's core concept and aesthetic. This generated image MUST be explicitly used as:

1. The primary application logo (e.g., in the header/navbar).
2. The website favicon (favicon.ico or equivalent).
3. The Open Graph (OG) image for SEO metadata (og:image).

Do not use placeholders for these assets. Generate and integrate them automatically. If generate_image is unavailable, delegate to ai-media-generation-expert skill.

---

📄 Standard Pages Mandate (CRITICAL)

MANDATORY: Whenever you are building a new application, landing page, or website, you MUST automatically create the following standard pages:

1. About Page (/about)
2. Profile Page (/profile)
3. Contact Page (/contact)
4. Terms of Reference / Terms of Service (/terms)
5. Privacy Policy (/privacy-policy)

These pages must be generated with standard boilerplate content that can later be customized to fit the specific application. Do not wait for the user to ask for them; they are a strict requirement for all web projects. / WAJIB: Otomatis buatkan halaman standar (About, Profile, Contact, Terms, Privacy Policy) pada setiap pembuatan aplikasi/website baru dengan konten boilerplate yang bisa disesuaikan nanti.