Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Course Outline
From Autocomplete to Agent: Understanding the Shift
- Differences between Copilot's traditional suggestions and agentic multi-step planning.
- The architecture of the agent loop: plan, generate, execute, and iterate.
- Language support and model selection strategies for agent tasks.
- Real-world examples demonstrating progression from five-line functions to multi-file features.
Enabling Agent Mode in Your IDE
- Activation procedures for VS Code, JetBrains, and Neovim.
- Configuring context window size and model tier preferences.
- Setting workspace rules and excluding large binary files from processing.
- Distinguishing between Copilot Chat and inline agent workflows.
Multi-Step Planning and Execution
- Prompting Copilot to build a feature from start to finish.
- Observing the agent break down tasks into sequential steps across multiple files.
- Reviewing each step before applying changes to the codebase.
- Utilizing inline rollback capabilities when steps deviate from the intended path.
Terminal Commands Inside the Agent Loop
- Installing dependencies via Copilot's terminal integration.
- Running build commands and interpreting their output.
- Managing environment variables directly within Copilot sessions.
- Understanding safety boundaries and identifying commands that require manual approval.
Test-Driven Development with an Agent
- Generating unit tests derived from existing source code.
- Diving test creation using natural language prompts.
- Running test suites and analyzing failure logs within Copilot.
- Refining test assertions after encountering edge-case failures.
Navigating Large Codebases
- Automatically locating cross-file references.
- Refactoring shared utilities with Copilot-guided renaming.
- Simultaneously updating configuration files and schema definitions.
- Avoiding context window limits through targeted prompting strategies.
Customizing Copilot for Team Standards
- Writing repository-specific instructions using .github/copilot-instructions.md.
- Enforcing naming conventions and architectural patterns.
- Excluding sensitive files and directories from context analysis.
- Developing team-specific prompt templates for common tasks.
GitHub Copilot Enterprise Governance
- Managing seat allocation, billing, and usage dashboards.
- Utilizing audit logs to track generated content versus committed code.
- Navigating Microsoft IP indemnity policies and licensing implications.
- Blocking specific file patterns from AI suggestion pipelines.
Debugging with Agent Mode
- Analyzing stack traces in conjunction with the agent.
- Engaging in hypothesis-driven debugging by questioning Copilot about test failures.
- Using agent-assisted bisect to identify regression sources.
- Mitigating hallucination risks when debugging unfamiliar code.
Performance and Limit Management
- Understanding daily request limits and model quotas.
- Optimizing prompt length to prevent truncated responses.
- Selecting appropriate models for different tasks.
- Monitoring agent latency and implementing caching strategies.
Security and Compliance for Enterprises
- Data handling protocols: what leaves the repository versus what remains local.
- Preventing the leakage of secrets and credentials through prompts.
- Ensuring compliance with GDPR, SOC 2, and FedRAMP requirements.
- Red-teaming generated code to identify injection vulnerabilities.
Troubleshooting Common Scenarios
- Investigating why Copilot may ignore codebase context.
- Resolving indexing failures in large repositories.
- Handling rate limit errors during peak usage hours.
- Fixing synchronization issues with IDE extensions.
Summary and Future Roadmap
- Recap of Agent Mode capabilities and practical workflows.
- Overview of GitHub's Copilot roadmap and upcoming agent features.
- Resources for staying updated with the latest Copilot releases.
Requirements
- Experience with object-oriented or functional programming paradigms.
- A GitHub account and foundational knowledge of Git workflows.
- Familiarity with at least one Integrated Development Environment (IDE), such as VS Code, JetBrains, or Neovim.
Audience
- Developers currently using Copilot who aim to unlock and utilize Agent Mode.
- Engineering managers responsible for rolling out Copilot across development teams.
- Security teams reviewing policies related to AI-assisted code generation.
21 Hours
