AI Prompts

A kit of instructions, prompts, and custom agents for GitHub Copilot and VS Code.

What Is This

Centralized repository containing:

• Agents (.github/agents/) – Specialized assistants with structured workflows

  • Planning Agent – Creates actionable plans with memory-bank persistence
  • Learning Coach – Interactive coding tutor with challenges and feedback

• Prompts (.github/prompts/) – Reusable commands

  • analyze, apply-pattern, document, refactor, review, test

• Instructions (.github/instructions/) – Best practices guides

  • Code quality, design patterns, documentation, error handling, security, git workflow, refactoring

• Copilot Instructions – Global configuration for VS Code

How to Use

1. Clone this repository or reference it in your workspace
2. Open in VS Code (instructions in .github/copilot-instructions.md are applied automatically)
3. Invoke agents: "Help me with Planning Agent" or use prompts directly

Structure

.github/
├── agents/          # Custom agents
├── prompts/         # Reusable prompts
├── instructions/    # Guides and best practices
└── copilot-instructions.md  # Global configuration

Contributing

Guidelines

• Maintain YAML frontmatter – Preserve --- markers and required fields (description, tools, name, etc.)
• Follow naming conventions – Use kebab-case for filenames (e.g., my-agent.agent.md, my-prompt.prompt.md)
• Document clearly – Include purpose, usage examples, and boundaries for new agents/prompts
• Reference existing patterns – Link to relevant instructions or similar agents
• Add new instructions – Document patterns in .github/instructions/ if they're broadly applicable

How to Contribute

1. Fork or clone this repository
2. Create a feature branch: git checkout -b feature/my-new-agent
3. Add your agent/prompt/instruction following the existing structure
4. Test locally by referencing the files in VS Code
5. Commit with clear messages: git commit -m "Add: new analysis agent for performance"
6. Push and open a pull request

What to Contribute

• New agents solving common problems (planning, research, refactoring, etc.)
• Prompts for specific workflows or domains
• Enhanced instructions based on team best practices
• Bug fixes or improvements to existing agents/prompts

License

MIT