Course Overview
This four-day workshop takes you from GitHub Copilot fundamentals to full agentic software engineering. Across twelve modules you will build expertise across the complete Copilot ecosystem, covering everything from artifacts and spec-driven planning to second brain workflows, MCP design, custom agents, plugins, agentic coding, the Copilot CLI and SDK, and AI-assisted DevOps. Every concept is grounded in hands-on practice with real tools and real scenarios, and examples span multiple technology stacks including Python, TypeScript, .NET, and Java so the patterns apply directly to the environment you already work in.
The way software gets built is changing fast. AI agents are no longer just productivity tricks; they are becoming core members of the engineering team. This course is designed for developers who want to lead that shift rather than catch up to it. You will learn to delegate the repetitive, the predictable, and the mechanical to agents so that you can focus your own attention on decisions that require judgment, context, and creativity. Exercises cover both greenfield projects where you define everything from scratch and brownfield codebases where you work with existing code, constraints, and legacy decisions.
You will start by mastering GitHub Copilot as a true development partner, then immediately ground that capability in spec-driven planning so every skill, agent, and workflow you build is purposeful. From there you will develop a second brain for continuous learning, design and consume MCP tools, implement custom agents, and orchestrate multi-agent systems. In the advanced modules you will use the Copilot CLI and SDK to automate at scale, build agent harnesses, and apply agentic patterns to DevOps. By the end you will have a working mental model for agentic systems, a set of reusable patterns you can bring back to your team, and the confidence to apply them immediately.
Who should attend
- Software engineers and developers looking to amplify productivity with AI agents
- Solution architects exploring AI-assisted design and code generation
- Technical leaders evaluating how agentic workflows enhance team velocity
- DevOps engineers applying AI to CI/CD, IaC, and cloud automation
Prerequisites
- Solid experience with software development and version control (Git/GitHub)
- Basic familiarity with cloud concepts (Azure or similar)
- A GitHub Copilot license is required — this is a bring-your-own-license class
Course Content
Module 1: GitHub Copilot Fundamentals
- GitHub Copilot Introduction
- Selecting Models
- AI Assisted Coding
- Slash Commands & Agent Controls
- The /init and /fork Slash Commands
- Context Variables
- Pull Requests & Code Reviews
- Management and Settings
Module 2: GitHub Copilot Artifacts & Tools
- Understanding and Shaping GitHub Copilot's Context Window
- Copilot Instructions & Prompt Files
- Model Context Protocol Essentials
- Skills & Custom Agents
- GitHub Copilot- & Agent-scoped Hooks
- Agent Debug Panel
- Maintaining and optimizing Copilot artifacts
- Setup GH Copilot in a greenfield / brownfield repo
Module 3: Planning & Spec-Driven Development
- Planning: Why It Matters in Agentic Development
- Planning Approaches
- Stepwise Planning
- Feedback Loop
- What are Specs
- Spec-Driven Development Benefits
- Spec Libraries Overview
Module 4: Creating & Managing Skills
- Skill Scoping: User, Repo, and Agent Level
- Adding Skills from Public Skill Resources
- Crafting Custom Skills
- Testing and Validating Skill Routing
- Composing Skills using Reference Skills
- Upgrading & Modernization Skills
Module 5: Second Brain & Continuous Learning
- Second Brain Concepts for Agentic Development
- Copilot Memory: User, Session and Repo Scope
- Hook-driven Knowledge Capture
- Agent Learning Loop
- Creating Skills from Conversation
- Enhancing Skills from recorded Conversations
- Automatic Skill Tuning
- Self-Improving Agent Workflows
Module 6: Consuming & Designing MCP Tools
- MCP Types: Remote, Local, Extensions & Custom
- Integrating existing MCP Servers
- Consuming Databases using MCP
- Implementing MCP Servers and wrapping REST APIs
- Designing MCP Apps
- MCP Authentication: OAuth, API Keys, and Tokens
- MCP Server Hosting and Deployment
- MCP Server Sandboxing
Module 7: Implementing Custom Agents
- Image & binary support for agents
- Agentic Browser Automation
- Docs Agent: Auto-Documentation and Commenting
- Frontend Agent: UI Code Generation from Design Specs
- AI Specialist Agent: Implementing a Domain-Specific Expert Agent
- Backend Agent: API Implementation from OpenAPI Specs
- Testing Agent: Generating Unit and Integration Tests
Module 8: Bundling & Sharing Artifacts using Plugins
- What is a GitHub Copilot Plugin?
- Plugin Structure & Manifest
- Bundling Skills, Hooks, and MCPs into Plugins
- Publishing and Sharing Plugins
- Installing and Managing Third-Party Plugins
Module 9: Implementing Agentic Coding
- Using Local Agents and Agent Mode
- Interrupting and Redirecting Running Agents
- Permissions & Autopilot Mode
- Delegating Tasks to Cloud Agents
- Using Background (Copilot CLI) Agents
- Agent Permissions & Approval Controls
- Multi Agent Orchestration & Hand-off Patterns
- Agent Harness Pattern
Module 10: GitHub Copilot CLI
- GitHub Copilot CLI
- Copilot CLI steering + permissions
- Running tasks autonomously and in parallel
- Business Case: HR Document Updates Automation
- Scripting Copilot CLI
- GitHub Agentic Workflows
Module 11: Copilot SDK & Automation
- GitHub Copilot SDK
- Session & Context Management
- Tool & Function Calling
- Custom Agent Composition
- Streaming Responses
- Testing Agentic Applications
- Building an Agent Harness with the Copilot SDK
Module 12: Agentic DevOps & Deployment
- Automation using Azure CLI
- Azure Developer CLI (azd) agentic mode
- Infrastructure as Code (IaC) - Bicep & Terraform
- Agentic Deployment using GitHub Actions
- Implementing a DevOps Agent for CI/CD Automation
English
German