Installation & Setup¶
Prerequisites¶
- VS Code (latest version)
- GitHub Copilot license (Individual, Business, or Enterprise)
- GitHub Copilot Chat extension
- Git
Clone the Repository¶
Open in VS Code¶
VS Code automatically recognizes copilot-instructions.md and loads the global context for Copilot.
Making Agents Available¶
Agents are located under agents/*.agent.md. GitHub Copilot automatically recognizes .agent.md files as agent definitions.
VS Code Workspace Trust
Make sure you trust the workspace so that Copilot can load the agent files.
Using Agents in Copilot Chat¶
Open Copilot Chat (Ctrl+Shift+I / Cmd+Shift+I) and reference an agent:
@workspace Use the Python Expert agent to refactor this service
using the Repository Pattern and Dependency Injection.
Selecting an Agent Directly¶
In VS Code, you can also select agents directly via the Copilot menu:
- Open Copilot Chat
- Click the agent dropdown
- Select the desired agent
Referencing Skills¶
Skills provide the agent with additional expertise. You can reference them explicitly:
@workspace Apply the Clean Code skill to improve this function.
Follow SOLID Principles and the Repository Pattern.
Project Structure¶
ai-agents/
├── agents/ # 15 specialized AI Agents
│ ├── lead-architect.agent.md
│ ├── python-expert.agent.md
│ ├── golang-expert.agent.md
│ └── ...
├── skills/ # Technical knowledge bases
│ ├── python-patterns/
│ ├── golang-patterns/
│ ├── software-engineering/
│ ├── architecture/
│ └── ...
├── marp-templates/ # Presentation templates
├── reference-repos/ # Reference architectures
├── copilot-instructions.md # Global Copilot context
└── mkdocs.yml # Documentation configuration
Next Steps¶
- Quick Start — Try your first agent
- Agent Catalog — Overview of all 18 agents
- Skills — Available knowledge bases