What problem does it solve? Making early architecture decisions for a new project is risky and time-consuming—choosing the wrong database, architecture style, or LLM provider can cost months of rework. This Skill provides a structured decision framework that walks you through requirements gathering, architecture pattern selection, data storage, caching, API design, resilience, and AI/LLM architecture in about 30 minutes. ## Core Features & Use Cases - Phased Decision Framework: Eight phases covering requirements collection, architecture style selection (Monolith vs Modular Monolith vs Microservices), database选型, caching strategies, API design, availability patterns, message queues, and AI/LLM architecture. - AI/LLM Architecture Guidance: Covers LLM provider selection (Claude, Gemini, Groq), RAG pipeline design, vector database选型 (pgvector, Pinecone, Weaviate), embedding models, prompt caching, and cost control. - Architecture Summary Output: Produces a structured architecture summary ready to be recorded as ADRs (Architecture Decision Records) via the companion /rsun-adr command. - Use Case: When starting a new client project, invoke this Skill to systematically decide whether to use PostgreSQL or MongoDB, whether you need Redis caching, which API style fits, and how to structure the AI provider fallback cascade—then record each decision as an ADR. ## Quick Start Ask the AI to help design the system architecture for your new project using the rsun-sysdesign framework, starting from your functional and non-functional requirements.
