System Architecture Design
Background & Applicable Scenarios
After the BA has clearly documented the requirements spec and business processes, the next question is: what tech stack should this system use? How many layers? What modules should it be split into? How is data stored? Where is it deployed? If these questions aren't answered before coding starts, the architecture will become increasingly chaotic. This skill uses BA outputs to help you design a production-ready system architecture step by step.
Input & Output
What you need to provide: Requirements Specification (PRD) + Business process diagrams, data flow diagrams, ER diagrams, application architecture diagrams (if already available).
What you will get: A standardized System Architecture Design Specification covering 7 architecture dimensions: Overall Architecture (architectural style & philosophy), Layered Architecture (layer responsibilities & technology selection), Module Decomposition (module responsibilities & dependencies), Interface Architecture (interface specifications & core interface definitions), Data Architecture (database/table design, caching strategy, data flow), Deployment Architecture (topology diagram & environment configuration), Non-functional Architecture (performance/security/high availability/observability solutions).
Process Description
The Agent first parses system goals, user roles, business domains, process nodes, data entities, and non-functional requirements from the PRD. It then searches the architecture knowledge base for reference solutions. Next, it designs dimension by dimension — each dimension's direction is confirmed before producing the solution: Overall Architecture → Layered Architecture → Module Decomposition → Interface Architecture → Data Architecture → Deployment Architecture → Non-functional Architecture. Finally, everything is integrated into a standardized System Architecture Design Specification.