Skip to content

config-schema.md

Latest commit

 

History

History
281 lines (231 loc) · 7.96 KB

config-schema.md

File metadata and controls

281 lines (231 loc) · 7.96 KB

CodingBuddy Configuration Schema

This document describes the configuration schema for CodingBuddy. The configuration file provides project context to AI assistants without overriding .ai-rules.

Quick Start

Create a codingbuddy.config.json file in your project root:

{
  "language": "ko",
  "projectName": "my-awesome-app",
  "techStack": {
    "frontend": ["React", "TypeScript"],
    "backend": ["NestJS"]
  }
}

Configuration File Formats

CodingBuddy supports only JSON configuration format:

  1. codingbuddy.config.json - JSON (only supported format)

Schema Reference

Basic Settings

Field Type Description Example
language string Communication language for AI responses (conversation only — all generated artifacts like code, comments, docs, and commit messages are always in English) 'ko', 'en', 'ja'
projectName string Project name 'my-app'
description string Project description 'E-commerce platform'
repository string Repository URL 'https://github.com/...'

Tech Stack (techStack)

Configure your project's technology stack:

{
  "techStack": {
    "languages": ["TypeScript", "Python"],
    "frontend": ["React", "Next.js", "Tailwind CSS"],
    "backend": ["NestJS", "Express"],
    "database": ["PostgreSQL", "Redis"],
    "infrastructure": ["Docker", "AWS", "Kubernetes"],
    "tools": ["ESLint", "Prettier", "Husky"],
    "details": {
      "TypeScript": {
        "version": "5.x",
        "notes": "Strict mode enabled"
      }
    }
  }
}

details is optional and provides deeper information about specific technologies.

Field Type Description
languages string[] Programming languages
frontend string[] Frontend frameworks/libraries
backend string[] Backend frameworks/libraries
database string[] Databases and data stores
infrastructure string[] Infrastructure/DevOps tools
tools string[] Development tools
details Record<string, TechDetail> (Deep) Detailed tech info

Architecture (architecture)

Define your project's architecture:

{
  "architecture": {
    "pattern": "feature-sliced",
    "structure": ["src/", "app/", "features/", "entities/", "shared/"],
    "componentStyle": "feature-based",
    "layers": [
      { "name": "app", "path": "src/app", "description": "Application layer" },
      { "name": "features", "path": "src/features", "dependencies": ["entities", "shared"] }
    ]
  }
}

layers is optional and provides detailed layer definitions with dependency information.

Field Type Description
pattern string Architecture pattern ('layered', 'hexagonal', 'clean', 'feature-sliced', etc.)
structure string[] Key directory paths
componentStyle string Component organization ('flat', 'grouped', 'feature-based')
layers ArchitectureLayer[] (Deep) Layer definitions

Conventions (conventions)

Specify coding conventions:

{
  "conventions": {
    "style": "airbnb",
    "naming": {
      "files": "kebab-case",
      "components": "PascalCase",
      "functions": "camelCase",
      "variables": "camelCase",
      "constants": "UPPER_SNAKE_CASE"
    },
    "importOrder": ["react", "@/", "~/", "."],
    "maxLineLength": 100,
    "semicolons": true,
    "quotes": "single",
    "rules": {
      "no-console": "warn"
    }
  }
}

rules is optional and allows custom linting rule overrides.

Field Type Description
style string Style guide ('airbnb', 'google', 'standard', 'prettier')
naming NamingConvention Naming convention rules
importOrder string[] Import ordering preference
maxLineLength number Maximum line length
semicolons boolean Use semicolons
quotes string Quote style ('single', 'double')
rules Record<string, unknown> (Deep) Custom linting rules

Test Strategy (testStrategy)

Configure testing approach:

{
  "testStrategy": {
    "approach": "tdd",
    "frameworks": ["vitest", "playwright"],
    "coverage": 80,
    "unitTestPattern": "colocated",
    "e2eDirectory": "e2e/",
    "mockingStrategy": "minimal"
  }
}
Field Type Description
approach string Testing approach ('tdd', 'bdd', 'test-after', 'mixed')
frameworks string[] Test frameworks
coverage number Target coverage percentage (0-100)
unitTestPattern string Unit test location ('colocated', 'separate')
e2eDirectory string E2E test directory
mockingStrategy string Mocking approach ('minimal', 'extensive', 'no-mocks')

AI Configuration (ai)

Configure AI model settings:

{
  "ai": {
    "defaultModel": "claude-sonnet-4-20250514"
  }
}
Field Type Description
defaultModel string Default Claude model for AI operations. Available options: claude-sonnet-4-20250514 (recommended, balanced), claude-opus-4-20250514 (most capable), claude-haiku-3-5-20241022 (fastest, not recommended)

Model Selection During Init:

When running npx codingbuddy init, you'll be prompted to select a default model:

  • Claude Sonnet 4 (Recommended) - Balanced performance and cost
  • Claude Opus 4 - Most capable, best for complex tasks
  • Claude Haiku 3.5 - Fastest, most cost-effective (not recommended for coding tasks)

Additional Context

{
  "keyFiles": [
    "src/core/types.ts",
    "docs/architecture.md",
    "CONTRIBUTING.md"
  ],
  "avoid": [
    "legacy-api",
    "deprecated-module"
  ],
  "custom": {
    "team": "Platform Team",
    "domain": "E-commerce"
  }
}
  • keyFiles: Key files AI should be aware of
  • avoid: Topics or areas to avoid
  • custom: Custom freeform context

Complete Example

{
  "language": "ko",
  "projectName": "wishket-platform",
  "description": "Freelancer marketplace platform",
  "repository": "https://github.com/example/wishket-platform",
  "techStack": {
    "languages": ["TypeScript"],
    "frontend": ["React", "Next.js", "Tailwind CSS", "React Query"],
    "backend": ["NestJS", "TypeORM"],
    "database": ["PostgreSQL", "Redis"],
    "infrastructure": ["Docker", "AWS ECS", "GitHub Actions"]
  },
  "architecture": {
    "pattern": "feature-sliced",
    "structure": ["src/app", "src/features", "src/entities", "src/shared"],
    "componentStyle": "feature-based"
  },
  "conventions": {
    "style": "airbnb",
    "naming": {
      "files": "kebab-case",
      "components": "PascalCase",
      "functions": "camelCase"
    },
    "quotes": "single",
    "semicolons": true
  },
  "testStrategy": {
    "approach": "tdd",
    "frameworks": ["vitest", "playwright"],
    "coverage": 80,
    "unitTestPattern": "colocated",
    "mockingStrategy": "minimal"
  },
  "ai": {
    "defaultModel": "claude-sonnet-4-20250514"
  },
  "keyFiles": ["src/shared/types/index.ts", "docs/api.md"],
  "avoid": ["legacy-v1-api"]
}

TypeScript Types

You can use TypeScript types for IDE autocomplete when authoring your JSON config:

import type { CodingBuddyConfig } from 'codingbuddy/config';

// Use this type to validate your codingbuddy.config.json structure
type Config = CodingBuddyConfig;

Validation

The configuration is validated at runtime using Zod schemas. Invalid configurations will produce clear error messages:

Error: Invalid configuration
  - techStack.coverage: Expected number, received string
  - conventions.quotes: Invalid enum value. Expected 'single' | 'double'

Notes

  • All fields are optional - start with what you need
  • No rule override - this config provides context only; .ai-rules rules remain unchanged
  • AI context only - this information helps AI understand your project better