AI Base Configuration Repository

Central repository for managing AI Instructions, Skills, and Agents that can be used by various AI systems.

🎯 Overview

This repository serves as a central source for:

• Instructions: General and language-specific coding guidelines
• Skills: Reusable capabilities for AI agents
• Agents: Specialized agent definitions for different roles

🏗️ Structure

ai-base/
├── src/
│   ├── instructions/         # Coding guidelines (*.instructions.md)
│   │   ├── general/          # General instructions
│   │   ├── csharp/           # C#-specific instructions
│   │   ├── typescript/       # TypeScript-specific instructions
│   │   ├── angular/          # Angular-specific instructions
│   │   ├── java/             # Java-specific instructions
│   │   └── docker/           # Docker-specific instructions
│   ├── skills/               # Reusable skills
│   │   ├── general/          # General skills (e.g., code-review)
│   │   ├── csharp/           # C#-specific skills (e.g., ef-core)
│   │   ├── typescript/       # TypeScript-specific skills (e.g., rxjs)
│   │   ├── angular/          # Angular-specific skills
│   │   └── java/             # Java-specific skills
│   └── agents/               # Agent definitions
│       ├── general/          # General agents (e.g., backend-developer)
│       ├── csharp/           # C#-specific agents
│       ├── typescript/       # TypeScript-specific agents
│       ├── angular/          # Angular-specific agents
│       └── java/             # Java-specific agents
├── tools/
│   ├── DeployAi/             # .NET deployment tool
│   └── deploy-ai-config.sh   # Deployment script (calls DeployAi)
└── .github/
    └── workflows/
        └── sync-ai-config.yml # Reusable GitHub Action

🚀 Usage

Using in Other Repositories

Create a workflow file in your repository (e.g., .github/workflows/sync-ai-config.yml):

name: Sync AI Configuration

on:
  workflow_dispatch:
    inputs:
      create-pull-request:
        description: 'Create a pull request instead of committing directly to the branch'
        required: false
        type: boolean
        default: true
      ai-base-version:
        description: 'Version/branch of ai-base to use'
        required: false
        type: string
        default: 'main'
  schedule:
    - cron: '0 0 * * *' # Every day at midnight

jobs:
  sync-scheduled:
    if: github.event_name == 'schedule'
    uses: CreativeCodersTeam/ai-base/.github/workflows/sync-ai-config.yml@main
    with:
      languages: 'csharp,typescript'
      ai-systems: 'copilot,claude,junie'
      ai-base-version: 'main'
      create-pull-request: true

  sync-manual:
    if: github.event_name == 'workflow_dispatch'
    uses: CreativeCodersTeam/ai-base/.github/workflows/sync-ai-config.yml@main
    with:
      languages: 'csharp,typescript'
      ai-systems: 'copilot,claude,junie'
      ai-base-version: ${{ inputs.ai-base-version }}
      create-pull-request: ${{ inputs.create-pull-request }}

Local Testing

# Deploy AI configuration using the DeployAi .NET program
sh ./tools/deploy-ai-config.sh \
  --languages=csharp,typescript \
  --ai-systems=copilot,junie,claude \
  --output-dir=./output

The deploy-ai-config.sh script invokes the DeployAi .NET program which generates and deploys the AI configuration files.

🤖 Supported AI Systems

GitHub Copilot

• Output files: .github/copilot-instructions.md
• Features: Combined instructions (as language-specific files are not supported)

Junie (JetBrains)

• Output files:
  • AGENTS.md (root)
  • .junie/guidelines.md
• Features: Combined instructions

Claude Code

• Output files:
  • CLAUDE.md (Instructions)
  • .claude/skills/ (Skills with SKILL.md files)
• Features:
  • Language-specific skills
  • Combined instructions

📝 Available Languages

• csharp - C# / .NET
• typescript - TypeScript
• angular - Angular Framework
• java - Java
• docker - Docker

⚙️ Parameters

GitHub Action Inputs

Parameter	Description	Required	Default
languages	Comma-separated list of languages	Yes	-
ai-systems	Comma-separated list of AI systems	Yes	-
ai-base-version	Version/branch of ai-base	No	main

Script Parameters

--languages=<lang1,lang2>    # Languages (comma-separated)
--ai-systems=<sys1,sys2>     # AI systems (comma-separated)
--output-dir=<path>          # Output directory (default: current directory)

📦 Adding New Content

Adding a New Instruction

1. Create a file matching the pattern *.instructions.md in src/instructions/<language>/
   • Example: coding-standards.instructions.md, best-practices.instructions.md
2. Follow the format of existing instructions
3. For language-specific instructions: Add **Scope: <Language> Projects**
4. All instruction files must follow the *.instructions.md naming pattern to be recognized by the deployment system

Adding a New Skill

1. Create a directory in src/skills/<language>/<skill-name>/
2. Add a SKILL.md file with:

   ---
   name: skill-name
   description: Description of when the skill should be used
   ---
   
   # Skill content

Adding a New Agent

1. Create a .md file in src/agents/<language>/
2. Define role, responsibilities, skills, and guidelines

🔄 How It Works

1. Source files: Instructions, Skills, and Agents are defined once in src/
2. Generator script: generate-ai-config.js reads the source files
3. Transformation:
   • For AI systems without language-specific support, files are combined
   • Headers are added with language scope
   • Files are converted to AI system-specific formats
4. Deployment: GitHub Action copies generated files to target repository

🎨 Combination Logic

For AI systems that don't support language-specific instructions:

1. General Instructions first

   # General Instructions
   <Content>

2. Language-specific Instructions after (for each language)

   # <LANGUAGE> Specific Instructions
   **Scope: <language> projects only**
   <Content>

📄 License

MIT

🤝 Contributing

Contributions are welcome! Please create a pull request with your changes.