ai-flow-dev 2.7.0 → 2.8.0

package/prompts/backend/flow-build-phase-3.md

@@ -1,834 +1,834 @@
-## PHASE 3: System Architecture (15-20 min)
-
-> **Order for this phase:** 3.1 → 3.2 → 3.3 → 3.4 → 3.5 → 3.6 → 3.7 → 3.8 → 3.9 → 3.10 → 3.11 → 3.12
-
-> **📌 Scope-based behavior:**
->
-> - **MVP:** Ask 3.1-3.6 (tech stack essentials) and 3.12 (API structure), skip 3.7-3.11 (advanced features), mark as "TBD"
-> - **Production-Ready:** Ask all questions 3.1-3.12
-> - **Enterprise:** Ask all questions 3.1-3.12 with emphasis on scalability and integrations
-
-> **📌 Note:** If Phase 0 detected framework/language/dependencies, those will be pre-filled. Review and confirm.
-
-### Objective
-
-Define the technical stack, architecture patterns, and system design.
-
-> **Note:** At the end of this phase, the AI will automatically generate a system architecture diagram in mermaid format, based on your answers. This diagram will be included in the docs/architecture.md document.
-
----
-
-## 🔍 Pre-Flight Check (Smart Skip Logic)
-
-> 📎 **Reference:** See [prompts/shared/smart-skip-preflight.md](../../.ai-flow/prompts/shared/smart-skip-preflight.md) for the complete smart skip logic.
-
-**Execute Pre-Flight Check for Phase 3:**
-
-- **Target File**: `docs/architecture.md`
-- **Phase Name**: "SYSTEM ARCHITECTURE"
-- **Key Items**: Framework, architecture pattern, API style, database, caching, background jobs, integrations
-- **Typical Gaps**: API versioning, rate limiting, caching strategy
-
-**Proceed with appropriate scenario based on audit data from `.ai-flow/cache/audit-data.json`**
-
----
-
-## Phase 3 Questions (Full Mode)
-
----
-
-#### 🎨 MERMAID ARCHITECTURE DIAGRAM FORMAT - CRITICAL
-
-> 📎 **Reference:** See [prompts/shared/mermaid-guidelines.md](../../.ai-flow/prompts/shared/mermaid-guidelines.md) for architecture diagram syntax, node shapes, and styling.
-
-**Example Architecture Diagram:**
-
-**Common Architecture Patterns:**
-
-```mermaid
-graph TD
-    subgraph "Client Layer"
-        Web[Web App]
-        Mobile[Mobile App]
-    end
-
-    subgraph "API Layer"
-        Gateway[API Gateway]
-        Auth[Auth Service]
-    end
-
-    subgraph "Business Layer"
-        Service1[User Service]
-        Service2[Order Service]
-        Service3[Payment Service]
-    end
-
-    subgraph "Data Layer"
-        DB[(PostgreSQL)]
-        Cache[(Redis)]
-    end
-
-    Web --> Gateway
-    Mobile --> Gateway
-    Gateway --> Auth
-    Gateway --> Service1
-    Gateway --> Service2
-    Service2 --> Service3
-    Service1 --> DB
-    Service2 --> DB
-    Service3 --> DB
-    Service1 --> Cache
-    Service2 --> Cache
-```
-
-**Best Practices:**
-
-- Group related components using `subgraph`
-- Show external services (Email, SMS, Payment gateways)
-- Include monitoring and logging components
-- Label protocols on connections (HTTPS, gRPC, WebSocket)
-- Use consistent naming conventions
-
-## **Validation:** Preview at https://mermaid.live/ before committing
-
-**3.1 Backend Framework**
-
-```
-[If detected from Phase 0, show:]
-✅ Framework Detected: [NestJS/FastAPI/Spring Boot/etc.]
-✅ Language: [TypeScript 5.3/Python 3.11/Java 21/etc.]
-✅ Runtime: [Node 20/Python 3.11/JVM 21/etc.]
-
-Is this correct? (Y/N)
-If no, please specify the correct framework and language.
-
-[If NOT detected, ask:]
-Which backend framework will you use?
-
-Node.js (JavaScript):
-A) 🔥 Express.js - Popular (minimal, flexible, lightweight)
-B) Hapi.js - Enterprise (configuration-driven)
-
-TypeScript (Node.js):
-C) ⭐ NestJS - Recommended (structured, enterprise-ready, decorators)
-D) ⚡ Fastify - Modern (high performance, schema validation)
-
-Python:
-E) ⭐ FastAPI - Recommended (modern, fast, auto-docs)
-F) 🔥 Django - Popular (batteries included, admin panel)
-G) Flask - Minimal (micro-framework, flexible)
-
-Java:
-H) 🏆 Spring Boot - Enterprise standard
-I) Quarkus - Modern (cloud-native, fast startup)
-
-Java (NetBeans - Ant Based):
-J) ⚡ NetBeans + Servlets/JSP - Traditional Java web
-   - Java EE web applications
-   - Servlet containers (Tomcat, GlassFish)
-   - JSP for server-side rendering
-   - JDBC for database access
-
-K) 🔥 NetBeans + JAX-RS - RESTful API
-   - Jersey or RESTEasy implementation
-   - JSON/XML response formats
-   - Bean Validation support
-   - GlassFish or Payara server
-
-L) 🏆 NetBeans + Spring Framework (Ant)
-   - Spring MVC or Spring Boot
-   - Dependency injection
-   - Can use Ant instead of Maven
-   - Tomcat embedded or standalone
-
-Java (Eclipse - Maven/Gradle):
-M) 🏆 Eclipse + Spring Boot (Maven/Gradle)
-   - Spring MVC or Spring REST
-   - Modern build tools
-   - Better dependency management
-   - Extensive Eclipse tooling
-
-Go:
-J) ⚡ Gin - Popular (fast, minimalist)
-K) Echo - Feature-rich (middleware, routing)
-L) Fiber - Express-like (high performance)
-
-Rust:
-M) ⚡ Actix-web - High performance (async, type-safe)
-N) Rocket - Developer-friendly (macros, type-safe)
-O) Axum - Modern (tokio-based, ergonomic)
-
-Kotlin:
-P) Ktor - Native Kotlin (coroutines, DSL)
-Q) Spring Boot - Java interop (Kotlin support)
-
-Other:
-R) Ruby (Rails)
-S) PHP (Laravel)
-T) C# (.NET Core)
-
-Your choice: __
-Why?
-```
-
-**3.2 Language & Version**
-
-```
-Primary programming language and version:
-
-Language: **
-Version: ** (e.g., Node 20, Python 3.11, Java 21)
-
-Type system:
-A) ⭐ Strongly typed - TypeScript, Java, Go (Recommended for large projects)
-B) Dynamically typed - JavaScript, Python, Ruby
-C) Gradually typed - Python with type hints
-
-Package Manager:
-A) ⭐ npm - Standard, comes with Node
-B) 🔥 pnpm - Fast, disk efficient
-C) ⚡ yarn - Popular alternative
-D) 🚀 bun - Ultra fast (if using Bun runtime)
-E) 🐍 pip/poetry (Python)
-F) ☕ Maven/Gradle (Java)
-G) 🐜 Apache Ant (NetBeans default, Java)
-
-**For NetBeans Projects:**
-
-Build Tool:
-A) ⭐ Apache Ant - Default for NetBeans (build.xml)
-B) Maven - Modern alternative (pom.xml)
-C) Gradle - Modern alternative (build.gradle)
-
-If using Ant:
-- Dependencies managed via lib/ folder
-- Manual JAR management or Ant + Ivy
-- Build targets: compile, jar, run, clean
-
-**For Eclipse Projects:**
-
-Build Tool:
-A) ⭐ Maven - Recommended (pom.xml)
-B) 🔥 Gradle - Modern alternative (build.gradle)
-C) Ant - Legacy (build.xml)
-
-Eclipse uses:
-- m2e (Maven integration)
-- Buildship (Gradle integration)
-- Automatic classpath management
-```
-
-**3.3 Architecture Pattern**
-
-```
-What architecture pattern will you follow?
-
-A) ⭐ Layered Architecture (Recommended for most projects)
-   - Presentation → Business Logic → Data Access
-   - Easy to understand and maintain
-
-B) 🏆 Hexagonal/Clean Architecture (Enterprise)
-   - Core domain isolated from infrastructure
-   - Highly testable and flexible
-
-C) 🔥 MVC (Popular, traditional)
-   - Model-View-Controller separation
-   - Good for traditional web apps
-
-D) 📦 Modular Monolith (Modern, scalable)
-   - Single deployment with independent modules
-   - Easier than microservices, more structured than monolith
-   - Good middle ground for growing applications
-
-E) ⚡ Microservices (Modern, complex)
-   - Multiple independent services
-   - Best for large-scale distributed systems
-
-F) Other: __
-
-Your choice: __
-Why this pattern?
-```
-
-**3.4 API Style**
-
-```
-What API style will you expose?
-
-A) ⭐ REST API - Recommended (HTTP/JSON, standard, well-understood)
-B) 🔥 GraphQL - Popular (flexible queries, single endpoint)
-C) ⚡ gRPC - Modern (high performance, protobuf, microservices)
-D) Mixed - REST + GraphQL or REST + gRPC
-
-Your choice: __
-
-API versioning strategy:
-A) URL versioning (/v1/users, /v2/users)
-B) Header versioning (Accept: application/vnd.api.v1+json)
-C) No versioning yet (will add when needed)
-```
-
-**3.5 API Reference (Automated)**
-
-````
-The AI will automatically generate standard CRUD endpoints for each entity defined in Phase 2.
-
-Please answer the following questions to define the global API conventions (these will apply to all endpoints unless otherwise specified):
-
-**A) Authentication and Access Control**
-1. Do all CRUD endpoints require authentication?
-  A) ⭐ Yes, all endpoints require authentication (recommended)
-  B) Only some (specify which ones)
-  C) No authentication required
-
-2. Which roles can access each CRUD operation?
-  - GET (list): [admin, manager, user]
-  - GET (detail): [admin, manager, user]
-  - POST (create): [admin, manager, user]
-  - PUT (update): [admin, manager]
-  - DELETE (delete): [admin]
-  (Standard example: admin, manager, user. Adjust as needed.)
-
-**B) Listing and Filter Conventions**
-3. Which pagination scheme do you prefer?
-  A) ⭐ offset/limit (recommended)
-  B) cursor-based
-  C) No pagination
-
-4. Which filter and sorting fields will be supported by default?
-  - Filters: [id, name, date, etc.]
-  - Sorting: [field, asc/desc]
-
-5. How will filters be passed for GET list endpoints?
-  A) ⭐ Query parameters (recommended for simple filters)
-     Example: GET /users?name=John&status=active&page=1&limit=10
-
-  B) POST /search endpoint with body (for complex filters)
-     Example: POST /users/search
-     Body: { "filters": { "name": "John", "status": "active" }, "page": 1, "limit": 10 }
-
-  C) Both (query params for simple, POST /search for complex)
-
-6. For POST/PUT/PATCH endpoints, will you use DTOs for request validation?
-  A) ⭐ Yes, strict DTOs with validation (recommended)
-  B) Accept raw JSON without strict schema
-
-  If yes, validation library: [from Phase 3.6 - class-validator, Zod, Pydantic, Joi]
-
-**C) Error and Response Structure**
-7. What error response format will be used?
-  A) Standard JSON:
-  ```json
-  {
-   "error": "Descriptive message",
-   "code": 400,
-   "details": {}
-  }
-  ```
-
-  B) Other (specify)
-
-8. Which fields will be included in the default successful response?
-  - data, meta (pagination), links, etc.
-
-**D) Relationships and Expansions**
-9. Allow expanding relationships (include/expand)?
-  A) ⭐ Yes, support `include` parameter (recommended)
-  B) No, flat data only
-
-**E) Custom Endpoint Example**
-10. If you want to customize an endpoint (e.g., add special logic, validations, or unique parameters), describe the case here:
-
-- [Brief description, example endpoint, parameters, special logic]
----
-The AI will use these conventions to automatically document all CRUD endpoints for each entity in api.md. If you need additional or custom endpoints, you can add them manually later.
-````
-
-**3.5.1 Error Codes Catalog**
-
-```
-Will you use standardized error codes?
-
-A) ⭐ Yes - Domain-specific error codes (recommended for APIs)
-B) No - HTTP status codes only
-
-If yes, define your error code format:
-
-Format:
-A) ⭐ Prefixed by domain: USER_001, ORDER_003, PAYMENT_005
-B) Numeric ranges: 1000-1999 (Users), 2000-2999 (Orders)
-C) Other: __
-
-Define your error codes:
-
-| Code          | HTTP | Message                        | Resolution                    |
-|---------------|------|--------------------------------|-------------------------------|
-| USER_001      | 404  | User not found                 | Verify user ID exists         |
-| USER_002      | 409  | Email already registered       | Use different email or login  |
-| USER_003      | 400  | Invalid email format           | Provide valid email           |
-| AUTH_001      | 401  | Invalid credentials            | Check username/password       |
-| AUTH_002      | 401  | Token expired                  | Refresh or re-authenticate    |
-| AUTH_003      | 403  | Insufficient permissions       | Contact administrator         |
-| ORDER_001     | 400  | Empty cart                     | Add items before checkout     |
-| ORDER_002     | 400  | Insufficient stock             | Reduce quantity or wait       |
-| PAYMENT_001   | 402  | Payment declined               | Try different payment method  |
-| VALIDATION_001| 400  | Required field missing         | Provide all required fields   |
-
-Your error codes:
-| Code | HTTP | Message | Resolution |
-|------|------|---------|------------|
-|      |      |         |            |
-```
-
-**3.5.2 Input Validation Rules Catalog**
-
-```
-Define validation rules for common fields across your API:
-
-| Field Type     | Rules                                    | Error Message                    |
-|----------------|------------------------------------------|----------------------------------|
-| email          | valid format, max 255, lowercase         | Invalid email format             |
-| password       | min 8, uppercase, lowercase, number      | Password too weak                |
-| username       | min 3, max 30, alphanumeric, no spaces   | Invalid username format          |
-| phone          | E.164 format or local format             | Invalid phone number             |
-| url            | valid URL, https only (optional)         | Invalid URL format               |
-| date           | ISO 8601 format, not in past (optional)  | Invalid date format              |
-| price/amount   | positive, max 2 decimals                 | Invalid amount                   |
-| quantity       | positive integer, max 9999               | Invalid quantity                 |
-| id (UUID)      | valid UUID v4 format                     | Invalid ID format                |
-| slug           | lowercase, hyphens only, max 100         | Invalid slug format              |
-
-Entity-specific validation (example):
-
-User:
-- firstName: required, min 2, max 50, letters only
-- lastName: required, min 2, max 50, letters only
-- birthDate: valid date, must be 18+ years ago
-
-Product:
-- name: required, min 3, max 100
-- price: required, positive, max 999999.99
-- sku: required, unique, uppercase, alphanumeric
-
-Your entity validations:
-
-Entity: __
-- field: [rules]
-
-Entity: __
-- field: [rules]
-```
-
-**3.5.3 Idempotency Strategy**
-
-```
-How will you handle duplicate requests (critical for payments, orders)?
-
-A) ⭐ Idempotency keys - Client sends unique key per request
-B) Natural idempotency - Use unique constraints (email, etc.)
-C) Not needed - Operations are naturally idempotent
-D) Combination of A + B
-
-If using idempotency keys (A):
-
-Header name:
-A) ⭐ Idempotency-Key (standard)
-B) X-Request-ID
-C) Custom: __
-
-Key storage:
-A) ⭐ Redis with TTL (recommended)
-B) Database table
-
-TTL: __ hours (recommended: 24)
-
-Which endpoints require idempotency?
-- POST /orders ✅
-- POST /payments ✅
-- POST /users ✅
-- [Your endpoints]: __
-```
-
-**3.6 Key Dependencies**
-
-```
-What major libraries/tools will you use?
-
-ORM/Database:
-A) TypeORM (Node.js)
-B) Prisma (Node.js) ⭐
-C) Sequelize (Node.js)
-D) SQLAlchemy (Python)
-E) Hibernate (Java)
-F) Other: __
-
-Validation:
-A) class-validator + class-transformer (NestJS) ⭐
-B) Joi (Node.js)
-C) Zod (TypeScript)
-D) Pydantic (Python) ⭐
-E) Yup (JavaScript)
-
-Authentication:
-A) Passport.js (Node.js) 🔥
-B) JWT libraries
-C) Auth0/Clerk/Supabase Auth (External service)
-D) Framework built-in
-
-Other critical libraries:
--
-```
-
-**3.7 Caching Strategy**
-
-```
-Will you use caching?
-
-A) ⭐ Redis - Recommended (in-memory, fast, pub/sub)
-B) Memcached - Simple key-value cache
-C) Application-level - In-process caching (node-cache, etc.)
-D) Database query cache
-E) No caching (simple projects)
-
-If using cache:
-- What will be cached? (sessions, query results, computed data)
-- Cache invalidation strategy? (TTL, manual, event-driven)
-```
-
-**3.8 Background Jobs**
-
-```
-Do you need background/async jobs?
-
-A) ⭐ Yes - Using queue system (Bull, BullMQ, Celery, Sidekiq)
-B) Yes - Using cron jobs
-C) Yes - Using serverless functions (Lambda, Cloud Functions)
-D) No - All operations are synchronous
-
-If yes, common job types:
-- Email sending
-- Report generation
-- Data processing
-- External API calls
-- Cleanup tasks
-- Other: __
-```
-
-**3.9 File Storage**
-
-```
-How will you handle file uploads?
-
-A) ⭐ Cloud storage - S3, Google Cloud Storage, Azure Blob ⭐
-B) Local filesystem - Storing on server disk
-C) Database - Storing binary data in DB (not recommended for large files)
-D) CDN - Cloudflare, CloudFront, etc.
-E) Not needed
-
-If storing files:
-- File types: [images, PDFs, videos, documents, etc.]
-- Max file size: __ MB
-- Storage quota estimate: __ GB
-```
-
-**3.10 API Gateway**
-
-```
-Will you use an API Gateway?
-
-A) ⭐ Yes - Using API Gateway (Kong, AWS API Gateway, Azure API Management, etc.)
-B) No - Direct API access
-
-If yes:
-- Gateway: __
-- Purpose: [Rate limiting, Authentication, Request routing, Load balancing, etc.]
-- Routes: __
-```
-
-**3.11 Real-time Communication**
-
-```
-Do you need real-time communication?
-
-A) ⭐ WebSockets - Bidirectional communication (chat, notifications, live updates)
-B) Server-Sent Events (SSE) - Server-to-client streaming (live feeds, updates)
-C) Both - Different use cases
-D) No - Standard HTTP requests only
-
-If WebSockets or SSE:
-- Use cases: __
-- Library: __
-- Authentication: __
-```
-
-**3.12 Message Broker Details** (if using background jobs from 3.8)
-
-```
-What message broker will you use?
-
-A) ⭐ RabbitMQ - Popular, reliable, feature-rich
-B) 🔥 Apache Kafka - High throughput, event streaming
-C) ⚡ AWS SQS - Managed, serverless
-D) Google Pub/Sub - Managed, scalable
-E) Redis Streams - Simple, fast
-F) Other: __
-
-Message patterns:
-A) ⭐ Queue - Point-to-point messaging
-B) Pub/Sub - Publish-subscribe pattern
-C) Both - Different use cases
-
-Delivery guarantees:
-A) ⭐ At-least-once - Messages delivered at least once (may have duplicates)
-B) Exactly-once - Messages delivered exactly once (more complex)
-C) At-most-once - Messages may be lost (rarely used)
-
-Dead letter queue:
-A) ⭐ Yes - Handle failed messages
-B) No
-```
-
-**3.13 API Documentation**
-
-```
-How will you document your API?
-
-A) ⭐ Swagger/OpenAPI - Auto-generated from code (code-first)
-   - Tool: [@nestjs/swagger, FastAPI docs, Swagger UI, etc.]
-   - Endpoint: /api-docs or /swagger
-
-B) 📝 OpenAPI Spec - Write spec first, generate code (design-first)
-   - File: openapi.yaml
-   - Tool: [OpenAPI Generator, etc.]
-
-C) Manual - Markdown documentation
-   - Not recommended (hard to keep in sync)
-
-Your choice: __
-```
-
-**3.14 Service Mesh** (if microservices architecture)
-
-```
-Will you use a Service Mesh?
-
-A) ⭐ Yes - Using Service Mesh (Istio, Linkerd, Consul Connect)
-B) No - Not needed (monolith or simple microservices)
-
-If yes:
-- Mesh: __
-- Features: [Service discovery, Load balancing, mTLS, Observability]
-```
-
-**3.15 External Integrations**
-
-```
-Will you integrate with external services?
-
-Select all that apply:
-
-💳 Payment Providers:
-A) Stripe - Credit cards, subscriptions ⭐
-B) PayPal - Popular payment method
-C) Square - POS and online payments
-D) Mercado Pago - Latin America
-E) Other: __
-
-→ Your selection (e.g., A): __
-
-📧 Email Services:
-A) AWS SES - Cost-effective, scalable ⭐
-B) SendGrid - Feature-rich, analytics
-C) Mailgun - Developer-friendly
-D) Postmark - Transactional focus
-E) Resend - Modern, simple API ⚡
-F) Other: __
-
-→ Your selection (e.g., A, B): __
-
-📱 SMS/Messaging:
-C) MessageBird - Multi-channel
-D) Other: __
-
-→ Your selection (e.g., A): __
-
-☁️ Cloud Storage:
-
-D) Cloudflare R2 - S3-compatible, no egress fees ⚡
-E) Other: __
-
-→ Your selection (e.g., A): __
-
-📊 Analytics: Storage
-D) Cloudflare R2 - S3-compatible, no egress fees ⚡
-E) Other: __
-
-📊 Analytics:
-E) Amplitude - Behavioral analytics
-F) Other: __
-
-→ Your selection (e.g., B, C): __
-
-🔍 Monitoring/Error Tracking:ytics ⚡
-E) Amplitude - Behavioral analytics
-D) LogRocket - Session replay
-E) Other: __
-
-→ Your selection (e.g., A): __
-
-🗺️ Maps/Location:tracking ⭐
-B) Datadog - Full observability 🏆
-C) New Relic - APM
-C) OpenStreetMap
-D) Other: __
-
-→ Your selection (e.g., A): __
-
-🔐 Authentication:
-A) Google Maps API
-D) Firebase Auth - Google ecosystem
-E) Other: __
-
-→ Your selection (e.g., A, B): __
-
-🤖 AI/ML Services:
-🔐 Authentication:
-D) AWS Bedrock - Managed AI
-E) Other: __
-
-→ Your selection (e.g., A): __
-
-📞 Communication:- Google ecosystem
-E) Other: __
-
-C) Webhooks - Custom integrations
-D) Other: __
-
-→ Your selection (e.g., A, B): __
-
-🔄 Other Integrations:timodal AI
-D) AWS Bedrock - Managed AI
-D) Accounting (QuickBooks, Xero)
-E) Other: __
-
-→ Your selection (e.g., A, B, C): __
----
-For each selected, briefly describe the use case:
-D) Other: __
-
-🔄 Other Integrations:
-A) GitHub/GitLab API
-B) Calendar (Google/Outlook)
-C) CRM (Salesforce, HubSpot)
-D) Accounting (QuickBooks, Xero)
-E) Other: __
----
-For each selected, briefly describe the use case:
-
-Example:
-- Stripe: Process credit card payments for subscriptions
-- AWS SES: Send transactional emails (order confirmations, password resets)
-- Sentry: Track and alert on production errors
-```
-
-### Phase 3 Output
-
-```
-📋 PHASE 3 SUMMARY:
-
-Framework: [name + version]
-Language: [name + version]
-Architecture: [pattern]
-API Style: [REST/GraphQL/gRPC]
-API Versioning: [strategy]
-API Conventions: [auth, pagination, error format, expansions]
-API Gateway: [yes/no + tool + purpose]
-Real-time Communication: [WebSockets/SSE/none + use cases]
-Message Broker: [tool + patterns + delivery guarantees]
-API Documentation: [Swagger/OpenAPI/manual + strategy]
-Service Mesh: [yes/no + tool if applicable]
-Database: [from Phase 2]
-ORM: [name]
-Validation: [library]
-Auth: [method]
-Caching: [strategy]
-Background Jobs: [yes/no + method]
-File Storage: [strategy]
-External Services: [list with use cases]
-
-Is this correct? (Yes/No)
-```
-
----
-
-### 📄 Generate Phase 3 Documents
-
-**Before starting generation:**
-
-```
-📖 Loading context from previous phases...
-✅ Re-reading project-brief.md
-✅ Re-reading docs/data-model.md
-```
-
-**Generate documents automatically:**
-
-**1. `docs/architecture.md`**
-
-- Use template: `.ai-flow/templates/docs/architecture.template.md`
-- Fill with system architecture, patterns, tech stack
-- Include architecture diagram (mermaid format)
-- Write to: `docs/architecture.md`
-
-**2. `ai-instructions.md`**
-
-- Use template: `.ai-flow/templates/ai-instructions.template.md`
-- Fill with tech stack, framework, language, key dependencies
-- Include NEVER/ALWAYS rules specific to chosen stack
-- Generate idiomatic code examples for Controller, Service, Repository, DTO and Module placeholders, strictly following the selected Architecture Pattern (e.g., if Hexagonal, show Ports & Adapters)
-- Write to: `ai-instructions.md`
-
-```
-✅ Generated: docs/architecture.md
-✅ Generated: ai-instructions.md
-
-Documents have been created with all Phase 3 information.
-
-📝 Would you like to make any corrections before continuing?
-
-→ If yes: Edit the files and type "ready" when done. I'll re-read them.
-→ If no: Type "continue" to proceed to Phase 4.
-```
-
-**If user edits files:**
-Execute `read_file()` for both documents to refresh context before continuing.
-
----
-
-**Proceed to Phase 4 only after documents are validated.**
-
-> ⚠️ **CRITICAL:** DO NOT generate README.md in this phase. README.md is ONLY generated in Phase 8 (step 8.5) after framework initialization.
-
----
-
-## 📝 Generated Documents
-
-After Phase 3, generate/update:
-
-- `docs/architecture.md` - Technical stack and patterns
-- `ai-instructions.md` - Instructions for AI agents
-
----
-
-**Next Phase:** Phase 4 - Security & Authentication (15-20 min)
-
-Read: `.ai-flow/prompts/backend/flow-build-phase-4.md`
-
----
-
-**Last Updated:** 2025-12-20
-**Version:** 2.1.8
-
----
-
-## PHASE 4: Security & Authentication (15-20 min)
+## PHASE 3: System Architecture (15-20 min)
+
+> **Order for this phase:** 3.1 → 3.2 → 3.3 → 3.4 → 3.5 → 3.6 → 3.7 → 3.8 → 3.9 → 3.10 → 3.11 → 3.12
+
+> **📌 Scope-based behavior:**
+>
+> - **MVP:** Ask 3.1-3.6 (tech stack essentials) and 3.12 (API structure), skip 3.7-3.11 (advanced features), mark as "TBD"
+> - **Production-Ready:** Ask all questions 3.1-3.12
+> - **Enterprise:** Ask all questions 3.1-3.12 with emphasis on scalability and integrations
+
+> **📌 Note:** If Phase 0 detected framework/language/dependencies, those will be pre-filled. Review and confirm.
+
+### Objective
+
+Define the technical stack, architecture patterns, and system design.
+
+> **Note:** At the end of this phase, the AI will automatically generate a system architecture diagram in mermaid format, based on your answers. This diagram will be included in the docs/architecture.md document.
+
+---
+
+## 🔍 Pre-Flight Check (Smart Skip Logic)
+
+> 📎 **Reference:** See [prompts/shared/smart-skip-preflight.md](../../.ai-flow/prompts/shared/smart-skip-preflight.md) for the complete smart skip logic.
+
+**Execute Pre-Flight Check for Phase 3:**
+
+- **Target File**: `docs/architecture.md`
+- **Phase Name**: "SYSTEM ARCHITECTURE"
+- **Key Items**: Framework, architecture pattern, API style, database, caching, background jobs, integrations
+- **Typical Gaps**: API versioning, rate limiting, caching strategy
+
+**Proceed with appropriate scenario based on audit data from `.ai-flow/cache/audit-data.json`**
+
+---
+
+## Phase 3 Questions (Full Mode)
+
+---
+
+#### 🎨 MERMAID ARCHITECTURE DIAGRAM FORMAT - CRITICAL
+
+> 📎 **Reference:** See [prompts/shared/mermaid-guidelines.md](../../.ai-flow/prompts/shared/mermaid-guidelines.md) for architecture diagram syntax, node shapes, and styling.
+
+**Example Architecture Diagram:**
+
+**Common Architecture Patterns:**
+
+```mermaid
+graph TD
+    subgraph "Client Layer"
+        Web[Web App]
+        Mobile[Mobile App]
+    end
+
+    subgraph "API Layer"
+        Gateway[API Gateway]
+        Auth[Auth Service]
+    end
+
+    subgraph "Business Layer"
+        Service1[User Service]
+        Service2[Order Service]
+        Service3[Payment Service]
+    end
+
+    subgraph "Data Layer"
+        DB[(PostgreSQL)]
+        Cache[(Redis)]
+    end
+
+    Web --> Gateway
+    Mobile --> Gateway
+    Gateway --> Auth
+    Gateway --> Service1
+    Gateway --> Service2
+    Service2 --> Service3
+    Service1 --> DB
+    Service2 --> DB
+    Service3 --> DB
+    Service1 --> Cache
+    Service2 --> Cache
+```
+
+**Best Practices:**
+
+- Group related components using `subgraph`
+- Show external services (Email, SMS, Payment gateways)
+- Include monitoring and logging components
+- Label protocols on connections (HTTPS, gRPC, WebSocket)
+- Use consistent naming conventions
+
+## **Validation:** Preview at https://mermaid.live/ before committing
+
+**3.1 Backend Framework**
+
+```
+[If detected from Phase 0, show:]
+✅ Framework Detected: [NestJS/FastAPI/Spring Boot/etc.]
+✅ Language: [TypeScript 5.3/Python 3.11/Java 21/etc.]
+✅ Runtime: [Node 20/Python 3.11/JVM 21/etc.]
+
+Is this correct? (Y/N)
+If no, please specify the correct framework and language.
+
+[If NOT detected, ask:]
+Which backend framework will you use?
+
+Node.js (JavaScript):
+A) 🔥 Express.js - Popular (minimal, flexible, lightweight)
+B) Hapi.js - Enterprise (configuration-driven)
+
+TypeScript (Node.js):
+C) ⭐ NestJS - Recommended (structured, enterprise-ready, decorators)
+D) ⚡ Fastify - Modern (high performance, schema validation)
+
+Python:
+E) ⭐ FastAPI - Recommended (modern, fast, auto-docs)
+F) 🔥 Django - Popular (batteries included, admin panel)
+G) Flask - Minimal (micro-framework, flexible)
+
+Java:
+H) 🏆 Spring Boot - Enterprise standard
+I) Quarkus - Modern (cloud-native, fast startup)
+
+Java (NetBeans - Ant Based):
+J) ⚡ NetBeans + Servlets/JSP - Traditional Java web
+   - Java EE web applications
+   - Servlet containers (Tomcat, GlassFish)
+   - JSP for server-side rendering
+   - JDBC for database access
+
+K) 🔥 NetBeans + JAX-RS - RESTful API
+   - Jersey or RESTEasy implementation
+   - JSON/XML response formats
+   - Bean Validation support
+   - GlassFish or Payara server
+
+L) 🏆 NetBeans + Spring Framework (Ant)
+   - Spring MVC or Spring Boot
+   - Dependency injection
+   - Can use Ant instead of Maven
+   - Tomcat embedded or standalone
+
+Java (Eclipse - Maven/Gradle):
+M) 🏆 Eclipse + Spring Boot (Maven/Gradle)
+   - Spring MVC or Spring REST
+   - Modern build tools
+   - Better dependency management
+   - Extensive Eclipse tooling
+
+Go:
+J) ⚡ Gin - Popular (fast, minimalist)
+K) Echo - Feature-rich (middleware, routing)
+L) Fiber - Express-like (high performance)
+
+Rust:
+M) ⚡ Actix-web - High performance (async, type-safe)
+N) Rocket - Developer-friendly (macros, type-safe)
+O) Axum - Modern (tokio-based, ergonomic)
+
+Kotlin:
+P) Ktor - Native Kotlin (coroutines, DSL)
+Q) Spring Boot - Java interop (Kotlin support)
+
+Other:
+R) Ruby (Rails)
+S) PHP (Laravel)
+T) C# (.NET Core)
+
+Your choice: __
+Why?
+```
+
+**3.2 Language & Version**
+
+```
+Primary programming language and version:
+
+Language: **
+Version: ** (e.g., Node 20, Python 3.11, Java 21)
+
+Type system:
+A) ⭐ Strongly typed - TypeScript, Java, Go (Recommended for large projects)
+B) Dynamically typed - JavaScript, Python, Ruby
+C) Gradually typed - Python with type hints
+
+Package Manager:
+A) ⭐ npm - Standard, comes with Node
+B) 🔥 pnpm - Fast, disk efficient
+C) ⚡ yarn - Popular alternative
+D) 🚀 bun - Ultra fast (if using Bun runtime)
+E) 🐍 pip/poetry (Python)
+F) ☕ Maven/Gradle (Java)
+G) 🐜 Apache Ant (NetBeans default, Java)
+
+**For NetBeans Projects:**
+
+Build Tool:
+A) ⭐ Apache Ant - Default for NetBeans (build.xml)
+B) Maven - Modern alternative (pom.xml)
+C) Gradle - Modern alternative (build.gradle)
+
+If using Ant:
+- Dependencies managed via lib/ folder
+- Manual JAR management or Ant + Ivy
+- Build targets: compile, jar, run, clean
+
+**For Eclipse Projects:**
+
+Build Tool:
+A) ⭐ Maven - Recommended (pom.xml)
+B) 🔥 Gradle - Modern alternative (build.gradle)
+C) Ant - Legacy (build.xml)
+
+Eclipse uses:
+- m2e (Maven integration)
+- Buildship (Gradle integration)
+- Automatic classpath management
+```
+
+**3.3 Architecture Pattern**
+
+```
+What architecture pattern will you follow?
+
+A) ⭐ Layered Architecture (Recommended for most projects)
+   - Presentation → Business Logic → Data Access
+   - Easy to understand and maintain
+
+B) 🏆 Hexagonal/Clean Architecture (Enterprise)
+   - Core domain isolated from infrastructure
+   - Highly testable and flexible
+
+C) 🔥 MVC (Popular, traditional)
+   - Model-View-Controller separation
+   - Good for traditional web apps
+
+D) 📦 Modular Monolith (Modern, scalable)
+   - Single deployment with independent modules
+   - Easier than microservices, more structured than monolith
+   - Good middle ground for growing applications
+
+E) ⚡ Microservices (Modern, complex)
+   - Multiple independent services
+   - Best for large-scale distributed systems
+
+F) Other: __
+
+Your choice: __
+Why this pattern?
+```
+
+**3.4 API Style**
+
+```
+What API style will you expose?
+
+A) ⭐ REST API - Recommended (HTTP/JSON, standard, well-understood)
+B) 🔥 GraphQL - Popular (flexible queries, single endpoint)
+C) ⚡ gRPC - Modern (high performance, protobuf, microservices)
+D) Mixed - REST + GraphQL or REST + gRPC
+
+Your choice: __
+
+API versioning strategy:
+A) URL versioning (/v1/users, /v2/users)
+B) Header versioning (Accept: application/vnd.api.v1+json)
+C) No versioning yet (will add when needed)
+```
+
+**3.5 API Reference (Automated)**
+
+````
+The AI will automatically generate standard CRUD endpoints for each entity defined in Phase 2.
+
+Please answer the following questions to define the global API conventions (these will apply to all endpoints unless otherwise specified):
+
+**A) Authentication and Access Control**
+1. Do all CRUD endpoints require authentication?
+  A) ⭐ Yes, all endpoints require authentication (recommended)
+  B) Only some (specify which ones)
+  C) No authentication required
+
+2. Which roles can access each CRUD operation?
+  - GET (list): [admin, manager, user]
+  - GET (detail): [admin, manager, user]
+  - POST (create): [admin, manager, user]
+  - PUT (update): [admin, manager]
+  - DELETE (delete): [admin]
+  (Standard example: admin, manager, user. Adjust as needed.)
+
+**B) Listing and Filter Conventions**
+3. Which pagination scheme do you prefer?
+  A) ⭐ offset/limit (recommended)
+  B) cursor-based
+  C) No pagination
+
+4. Which filter and sorting fields will be supported by default?
+  - Filters: [id, name, date, etc.]
+  - Sorting: [field, asc/desc]
+
+5. How will filters be passed for GET list endpoints?
+  A) ⭐ Query parameters (recommended for simple filters)
+     Example: GET /users?name=John&status=active&page=1&limit=10
+
+  B) POST /search endpoint with body (for complex filters)
+     Example: POST /users/search
+     Body: { "filters": { "name": "John", "status": "active" }, "page": 1, "limit": 10 }
+
+  C) Both (query params for simple, POST /search for complex)
+
+6. For POST/PUT/PATCH endpoints, will you use DTOs for request validation?
+  A) ⭐ Yes, strict DTOs with validation (recommended)
+  B) Accept raw JSON without strict schema
+
+  If yes, validation library: [from Phase 3.6 - class-validator, Zod, Pydantic, Joi]
+
+**C) Error and Response Structure**
+7. What error response format will be used?
+  A) Standard JSON:
+  ```json
+  {
+   "error": "Descriptive message",
+   "code": 400,
+   "details": {}
+  }
+  ```
+
+  B) Other (specify)
+
+8. Which fields will be included in the default successful response?
+  - data, meta (pagination), links, etc.
+
+**D) Relationships and Expansions**
+9. Allow expanding relationships (include/expand)?
+  A) ⭐ Yes, support `include` parameter (recommended)
+  B) No, flat data only
+
+**E) Custom Endpoint Example**
+10. If you want to customize an endpoint (e.g., add special logic, validations, or unique parameters), describe the case here:
+
+- [Brief description, example endpoint, parameters, special logic]
+---
+The AI will use these conventions to automatically document all CRUD endpoints for each entity in api.md. If you need additional or custom endpoints, you can add them manually later.
+````
+
+**3.5.1 Error Codes Catalog**
+
+```
+Will you use standardized error codes?
+
+A) ⭐ Yes - Domain-specific error codes (recommended for APIs)
+B) No - HTTP status codes only
+
+If yes, define your error code format:
+
+Format:
+A) ⭐ Prefixed by domain: USER_001, ORDER_003, PAYMENT_005
+B) Numeric ranges: 1000-1999 (Users), 2000-2999 (Orders)
+C) Other: __
+
+Define your error codes:
+
+| Code          | HTTP | Message                        | Resolution                    |
+|---------------|------|--------------------------------|-------------------------------|
+| USER_001      | 404  | User not found                 | Verify user ID exists         |
+| USER_002      | 409  | Email already registered       | Use different email or login  |
+| USER_003      | 400  | Invalid email format           | Provide valid email           |
+| AUTH_001      | 401  | Invalid credentials            | Check username/password       |
+| AUTH_002      | 401  | Token expired                  | Refresh or re-authenticate    |
+| AUTH_003      | 403  | Insufficient permissions       | Contact administrator         |
+| ORDER_001     | 400  | Empty cart                     | Add items before checkout     |
+| ORDER_002     | 400  | Insufficient stock             | Reduce quantity or wait       |
+| PAYMENT_001   | 402  | Payment declined               | Try different payment method  |
+| VALIDATION_001| 400  | Required field missing         | Provide all required fields   |
+
+Your error codes:
+| Code | HTTP | Message | Resolution |
+|------|------|---------|------------|
+|      |      |         |            |
+```
+
+**3.5.2 Input Validation Rules Catalog**
+
+```
+Define validation rules for common fields across your API:
+
+| Field Type     | Rules                                    | Error Message                    |
+|----------------|------------------------------------------|----------------------------------|
+| email          | valid format, max 255, lowercase         | Invalid email format             |
+| password       | min 8, uppercase, lowercase, number      | Password too weak                |
+| username       | min 3, max 30, alphanumeric, no spaces   | Invalid username format          |
+| phone          | E.164 format or local format             | Invalid phone number             |
+| url            | valid URL, https only (optional)         | Invalid URL format               |
+| date           | ISO 8601 format, not in past (optional)  | Invalid date format              |
+| price/amount   | positive, max 2 decimals                 | Invalid amount                   |
+| quantity       | positive integer, max 9999               | Invalid quantity                 |
+| id (UUID)      | valid UUID v4 format                     | Invalid ID format                |
+| slug           | lowercase, hyphens only, max 100         | Invalid slug format              |
+
+Entity-specific validation (example):
+
+User:
+- firstName: required, min 2, max 50, letters only
+- lastName: required, min 2, max 50, letters only
+- birthDate: valid date, must be 18+ years ago
+
+Product:
+- name: required, min 3, max 100
+- price: required, positive, max 999999.99
+- sku: required, unique, uppercase, alphanumeric
+
+Your entity validations:
+
+Entity: __
+- field: [rules]
+
+Entity: __
+- field: [rules]
+```
+
+**3.5.3 Idempotency Strategy**
+
+```
+How will you handle duplicate requests (critical for payments, orders)?
+
+A) ⭐ Idempotency keys - Client sends unique key per request
+B) Natural idempotency - Use unique constraints (email, etc.)
+C) Not needed - Operations are naturally idempotent
+D) Combination of A + B
+
+If using idempotency keys (A):
+
+Header name:
+A) ⭐ Idempotency-Key (standard)
+B) X-Request-ID
+C) Custom: __
+
+Key storage:
+A) ⭐ Redis with TTL (recommended)
+B) Database table
+
+TTL: __ hours (recommended: 24)
+
+Which endpoints require idempotency?
+- POST /orders ✅
+- POST /payments ✅
+- POST /users ✅
+- [Your endpoints]: __
+```
+
+**3.6 Key Dependencies**
+
+```
+What major libraries/tools will you use?
+
+ORM/Database:
+A) TypeORM (Node.js)
+B) Prisma (Node.js) ⭐
+C) Sequelize (Node.js)
+D) SQLAlchemy (Python)
+E) Hibernate (Java)
+F) Other: __
+
+Validation:
+A) class-validator + class-transformer (NestJS) ⭐
+B) Joi (Node.js)
+C) Zod (TypeScript)
+D) Pydantic (Python) ⭐
+E) Yup (JavaScript)
+
+Authentication:
+A) Passport.js (Node.js) 🔥
+B) JWT libraries
+C) Auth0/Clerk/Supabase Auth (External service)
+D) Framework built-in
+
+Other critical libraries:
+-
+```
+
+**3.7 Caching Strategy**
+
+```
+Will you use caching?
+
+A) ⭐ Redis - Recommended (in-memory, fast, pub/sub)
+B) Memcached - Simple key-value cache
+C) Application-level - In-process caching (node-cache, etc.)
+D) Database query cache
+E) No caching (simple projects)
+
+If using cache:
+- What will be cached? (sessions, query results, computed data)
+- Cache invalidation strategy? (TTL, manual, event-driven)
+```
+
+**3.8 Background Jobs**
+
+```
+Do you need background/async jobs?
+
+A) ⭐ Yes - Using queue system (Bull, BullMQ, Celery, Sidekiq)
+B) Yes - Using cron jobs
+C) Yes - Using serverless functions (Lambda, Cloud Functions)
+D) No - All operations are synchronous
+
+If yes, common job types:
+- Email sending
+- Report generation
+- Data processing
+- External API calls
+- Cleanup tasks
+- Other: __
+```
+
+**3.9 File Storage**
+
+```
+How will you handle file uploads?
+
+A) ⭐ Cloud storage - S3, Google Cloud Storage, Azure Blob ⭐
+B) Local filesystem - Storing on server disk
+C) Database - Storing binary data in DB (not recommended for large files)
+D) CDN - Cloudflare, CloudFront, etc.
+E) Not needed
+
+If storing files:
+- File types: [images, PDFs, videos, documents, etc.]
+- Max file size: __ MB
+- Storage quota estimate: __ GB
+```
+
+**3.10 API Gateway**
+
+```
+Will you use an API Gateway?
+
+A) ⭐ Yes - Using API Gateway (Kong, AWS API Gateway, Azure API Management, etc.)
+B) No - Direct API access
+
+If yes:
+- Gateway: __
+- Purpose: [Rate limiting, Authentication, Request routing, Load balancing, etc.]
+- Routes: __
+```
+
+**3.11 Real-time Communication**
+
+```
+Do you need real-time communication?
+
+A) ⭐ WebSockets - Bidirectional communication (chat, notifications, live updates)
+B) Server-Sent Events (SSE) - Server-to-client streaming (live feeds, updates)
+C) Both - Different use cases
+D) No - Standard HTTP requests only
+
+If WebSockets or SSE:
+- Use cases: __
+- Library: __
+- Authentication: __
+```
+
+**3.12 Message Broker Details** (if using background jobs from 3.8)
+
+```
+What message broker will you use?
+
+A) ⭐ RabbitMQ - Popular, reliable, feature-rich
+B) 🔥 Apache Kafka - High throughput, event streaming
+C) ⚡ AWS SQS - Managed, serverless
+D) Google Pub/Sub - Managed, scalable
+E) Redis Streams - Simple, fast
+F) Other: __
+
+Message patterns:
+A) ⭐ Queue - Point-to-point messaging
+B) Pub/Sub - Publish-subscribe pattern
+C) Both - Different use cases
+
+Delivery guarantees:
+A) ⭐ At-least-once - Messages delivered at least once (may have duplicates)
+B) Exactly-once - Messages delivered exactly once (more complex)
+C) At-most-once - Messages may be lost (rarely used)
+
+Dead letter queue:
+A) ⭐ Yes - Handle failed messages
+B) No
+```
+
+**3.13 API Documentation**
+
+```
+How will you document your API?
+
+A) ⭐ Swagger/OpenAPI - Auto-generated from code (code-first)
+   - Tool: [@nestjs/swagger, FastAPI docs, Swagger UI, etc.]
+   - Endpoint: /api-docs or /swagger
+
+B) 📝 OpenAPI Spec - Write spec first, generate code (design-first)
+   - File: openapi.yaml
+   - Tool: [OpenAPI Generator, etc.]
+
+C) Manual - Markdown documentation
+   - Not recommended (hard to keep in sync)
+
+Your choice: __
+```
+
+**3.14 Service Mesh** (if microservices architecture)
+
+```
+Will you use a Service Mesh?
+
+A) ⭐ Yes - Using Service Mesh (Istio, Linkerd, Consul Connect)
+B) No - Not needed (monolith or simple microservices)
+
+If yes:
+- Mesh: __
+- Features: [Service discovery, Load balancing, mTLS, Observability]
+```
+
+**3.15 External Integrations**
+
+```
+Will you integrate with external services?
+
+Select all that apply:
+
+💳 Payment Providers:
+A) Stripe - Credit cards, subscriptions ⭐
+B) PayPal - Popular payment method
+C) Square - POS and online payments
+D) Mercado Pago - Latin America
+E) Other: __
+
+→ Your selection (e.g., A): __
+
+📧 Email Services:
+A) AWS SES - Cost-effective, scalable ⭐
+B) SendGrid - Feature-rich, analytics
+C) Mailgun - Developer-friendly
+D) Postmark - Transactional focus
+E) Resend - Modern, simple API ⚡
+F) Other: __
+
+→ Your selection (e.g., A, B): __
+
+📱 SMS/Messaging:
+C) MessageBird - Multi-channel
+D) Other: __
+
+→ Your selection (e.g., A): __
+
+☁️ Cloud Storage:
+
+D) Cloudflare R2 - S3-compatible, no egress fees ⚡
+E) Other: __
+
+→ Your selection (e.g., A): __
+
+📊 Analytics: Storage
+D) Cloudflare R2 - S3-compatible, no egress fees ⚡
+E) Other: __
+
+📊 Analytics:
+E) Amplitude - Behavioral analytics
+F) Other: __
+
+→ Your selection (e.g., B, C): __
+
+🔍 Monitoring/Error Tracking:ytics ⚡
+E) Amplitude - Behavioral analytics
+D) LogRocket - Session replay
+E) Other: __
+
+→ Your selection (e.g., A): __
+
+🗺️ Maps/Location:tracking ⭐
+B) Datadog - Full observability 🏆
+C) New Relic - APM
+C) OpenStreetMap
+D) Other: __
+
+→ Your selection (e.g., A): __
+
+🔐 Authentication:
+A) Google Maps API
+D) Firebase Auth - Google ecosystem
+E) Other: __
+
+→ Your selection (e.g., A, B): __
+
+🤖 AI/ML Services:
+🔐 Authentication:
+D) AWS Bedrock - Managed AI
+E) Other: __
+
+→ Your selection (e.g., A): __
+
+📞 Communication:- Google ecosystem
+E) Other: __
+
+C) Webhooks - Custom integrations
+D) Other: __
+
+→ Your selection (e.g., A, B): __
+
+🔄 Other Integrations:timodal AI
+D) AWS Bedrock - Managed AI
+D) Accounting (QuickBooks, Xero)
+E) Other: __
+
+→ Your selection (e.g., A, B, C): __
+---
+For each selected, briefly describe the use case:
+D) Other: __
+
+🔄 Other Integrations:
+A) GitHub/GitLab API
+B) Calendar (Google/Outlook)
+C) CRM (Salesforce, HubSpot)
+D) Accounting (QuickBooks, Xero)
+E) Other: __
+---
+For each selected, briefly describe the use case:
+
+Example:
+- Stripe: Process credit card payments for subscriptions
+- AWS SES: Send transactional emails (order confirmations, password resets)
+- Sentry: Track and alert on production errors
+```
+
+### Phase 3 Output
+
+```
+📋 PHASE 3 SUMMARY:
+
+Framework: [name + version]
+Language: [name + version]
+Architecture: [pattern]
+API Style: [REST/GraphQL/gRPC]
+API Versioning: [strategy]
+API Conventions: [auth, pagination, error format, expansions]
+API Gateway: [yes/no + tool + purpose]
+Real-time Communication: [WebSockets/SSE/none + use cases]
+Message Broker: [tool + patterns + delivery guarantees]
+API Documentation: [Swagger/OpenAPI/manual + strategy]
+Service Mesh: [yes/no + tool if applicable]
+Database: [from Phase 2]
+ORM: [name]
+Validation: [library]
+Auth: [method]
+Caching: [strategy]
+Background Jobs: [yes/no + method]
+File Storage: [strategy]
+External Services: [list with use cases]
+
+Is this correct? (Yes/No)
+```
+
+---
+
+### 📄 Generate Phase 3 Documents
+
+**Before starting generation:**
+
+```
+📖 Loading context from previous phases...
+✅ Re-reading project-brief.md
+✅ Re-reading docs/data-model.md
+```
+
+**Generate documents automatically:**
+
+**1. `docs/architecture.md`**
+
+- Use template: `.ai-flow/templates/docs/architecture.template.md`
+- Fill with system architecture, patterns, tech stack
+- Include architecture diagram (mermaid format)
+- Write to: `docs/architecture.md`
+
+**2. `ai-instructions.md`**
+
+- Use template: `.ai-flow/templates/ai-instructions.template.md`
+- Fill with tech stack, framework, language, key dependencies
+- Include NEVER/ALWAYS rules specific to chosen stack
+- Generate idiomatic code examples for Controller, Service, Repository, DTO and Module placeholders, strictly following the selected Architecture Pattern (e.g., if Hexagonal, show Ports & Adapters)
+- Write to: `ai-instructions.md`
+
+```
+✅ Generated: docs/architecture.md
+✅ Generated: ai-instructions.md
+
+Documents have been created with all Phase 3 information.
+
+📝 Would you like to make any corrections before continuing?
+
+→ If yes: Edit the files and type "ready" when done. I'll re-read them.
+→ If no: Type "continue" to proceed to Phase 4.
+```
+
+**If user edits files:**
+Execute `read_file()` for both documents to refresh context before continuing.
+
+---
+
+**Proceed to Phase 4 only after documents are validated.**
+
+> ⚠️ **CRITICAL:** DO NOT generate README.md in this phase. README.md is ONLY generated in Phase 8 (step 8.5) after framework initialization.
+
+---
+
+## 📝 Generated Documents
+
+After Phase 3, generate/update:
+
+- `docs/architecture.md` - Technical stack and patterns
+- `ai-instructions.md` - Instructions for AI agents
+
+---
+
+**Next Phase:** Phase 4 - Security & Authentication (15-20 min)
+
+Read: `.ai-flow/prompts/backend/flow-build-phase-4.md`
+
+---
+
+**Last Updated:** 2025-12-20
+**Version:** 2.1.8
+
+---
+
+## PHASE 4: Security & Authentication (15-20 min)