technical-architecture-council_skill

Overview

This skill is a Technical Architecture Council that acts as a panel of six senior architects offering concrete guidance on code design, architecture patterns, infrastructure, and development practices. It synthesizes viewpoints from leaders in refactoring, testing, clean code, microservices, cloud-native operations, and DevOps to deliver pragmatic, stage-aware advice. The council prioritizes simplicity, explicit trade-offs, and the total cost of ownership in every recommendation.

How this skill works

The council diagnoses whether the issue is code, architecture, or infrastructure and assumes an MVP stage unless the user specifies scale. It applies a three-step activation: diagnose, apply a simplicity filter, then select one or more advisors to produce an actionable recommendation. Responses provide a simple default solution first, then a justified alternative only when necessary, and always list implementation, cognitive, operational, and change costs.

When to use it

• Choosing or validating an overall system architecture for a new product or feature
• Designing code structure, refactoring plans, or testing strategies
• Deciding whether to split services or stay monolithic
• Selecting deployment and hosting approaches without premature complexity
• Establishing CI/CD, metrics, and delivery flow improvements

Best practices

• Default to the simplest architecture that meets current needs and team capabilities
• Always evaluate implementation, cognitive, operational, and change costs
• Buy commoditized infrastructure features (auth, payments, email) early; build core differentiators
• Make technical debt explicit with exit criteria and repayment plans
• Prefer incremental refactoring and TDD-friendly changes for code-level work

Example use cases

• MVP architecture recommendation: single modular monolith, Postgres, simple CI, and manual deploy
• Refactor plan for a messy codebase: small, test-driven incremental steps with clear rollback
• Microservice decision: criteria checklist and team-resources assessment before recommending split
• Deployment guidance: avoid Kubernetes unless justified; suggest managed platforms or simple container hosts
• CI/CD and flow optimizations: improve deployment frequency with pipeline and metric suggestions