Standardization of Boilerplate Templates Integrating AI Coding Guidelines and Skills

The platform engineering team standardized boilerplate templates to reduce the repetitive development and operational burden of new projects. Previously, a mix of tech stacks, inconsistent conventions, and duplicate implementations of foundational features such as authentication and authorization made it difficult to achieve rapid releases and consistency.

The new template is based on a validated common structure but excludes business logic, standardizing only foundational features such as authentication, authorization, logging, and build. It provides structured example code along with AI coding guidelines (rules), designed to enable immediate development simply by changing the service name and domain.

Architecture and Tech Stack

The backend is based on Kotlin·Spring Boot, separating the application core layer where business logic is concentrated from different entry points such as API, Bot, MCP, and Batch. The frontend uses TypeScript·React and includes Antd-based UI and common components.

• Backend: OIDC-based JWT validation, Spring Data JPA/Kotlin JDSL, Redis, Actuator, Swagger applied
• Frontend: Antd UI, OIDC login, menu·permission management, husky/eslint/xo conventions applied
• AI folder: Includes coding guidelines (rules) and executable capabilities (skills) for analysis·deployment, etc.

Role of AI Coding Guidelines and Skills

The core differentiator is standardizing the way AI generates code itself. AI coding guidelines codify architecture, conventions, security, and testing rules, forcing both humans and AI to generate code of the same quality. Unlike traditional linters that check written code, this is fundamental standardization that aligns the code generation process itself.

It also includes built-in AI Skills for delegating repetitive tasks. These provide features such as project analysis, deployment, log·system analysis, code convention review, and security review, supporting developers in focusing on solving business problems.

Scalability and Future Plans

Business logic is implemented only in the Application layer, while API·Bot·MCP·Batch serve only as adapters calling it, preventing logic duplication. In the future, the team plans to standardize internal system integration via MCP servers and expose them through Bots, allowing standard tasks such as deployment or log analysis to be invoked from messenger channels.