Training program

Copilot as a Developer Workflow Partner

1. Copilot as a code generator and a tool for understanding systems
2. Working with existing code as the primary use case
3. Work quality, developer responsibility, and model limitations
4. The role of AI in analysis, implementation, testing, and review

Project Context in VS Code

1. Workspace, open files, selection, symbols, terminal, and errors
2. Controlling the scope of context passed to Copilot
3. Documentation, tests, and active files as model signals
4. The impact of context on answer quality and precision

GitHub Copilot Working Modes

1. Autocomplete and the nearest editor context
2. Inline Chat as a tool for small local changes
3. Chat as support for analysis, design, and comparing options
4. Agent Mode under control: plan, scope, and tests as the source of truth

AI Workflow Frameworks in a Legacy Codebase

1. Goal -> Context -> Constraints -> Ask -> Verify -> Iterate
2. Understand -> Scope -> Patch -> Test -> Review
3. Explain -> Challenge -> Improve
4. Bad prompt -> Better context -> Smaller task -> Verified result
5. Narrowing the request and controlling change risk

Workshop Project

1. Working with a sample existing codebase
2. Modules across different layers or technologies
3. Partially outdated documentation
4. Legacy smells, hidden dependencies, and missing tests
5. Analysis of project structure, data flow, and risk points
6. Task cards adapted to the selected workshop variant

Context Engineering in Practice

1. Packaging context for AI-assisted work
2. Working with files, symbols, tests, and documentation
3. Small scopes of change and explicit implementation constraints
4. Asking for a plan before implementation
5. Preparing context for a parser module change

Characterization Tests and Golden Master Testing

1. Tests that describe current system behavior
2. Red -> Green -> Refactor with AI
3. Golden master as protection against unintended behavior changes
4. Working with input files and expected reports
5. Deliberately detecting output changes

Safe Refactoring with AI

1. Refactoring scope and contract protection
2. Working in small steps
3. Verification through tests and golden output
4. Reviewing diffs generated with AI support
5. Separating refactoring from behavior changes

Debugging and Root Cause Analysis

1. Stack traces, logs, regressions, and incorrect reports
2. Explain -> Challenge -> Improve
3. Copilot as a tool for building hypotheses
4. Verifying causes in code, tests, and input data

Testing and Evaluating AI Suggestions

1. Generating tests with Copilot
2. Identifying missing scenarios
3. False confidence in model answers
4. Checklist for evaluating Copilot responses
5. Reviewing risky AI suggestions

Security, Secrets, and Licenses

1. Sensitive data in context
2. Secrets, configuration, and realistic test data
3. Public code matching
4. Reviewing changes for security and licensing risk

Repository Instructions and Reusable Prompts

1. `.github/copilot-instructions.md`
2. Prompt files
3. Instructions as team memory
4. Good and bad uses of repository instructions
5. Prompts that support module analysis and change review

GitHub Workflow, Copilot CLI, Coding Agent, and MCP

1. Local work in VS Code as the main workflow
2. Copilot in the terminal
3. Copilot Coding Agent
4. GitHub MCP in the repository context
5. Evaluating AI tool output in the team workflow