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 Paradigm Shift
- Distinguishing Copilot suggestions from agentic multi-step planning.
- The architecture of the agent loop: plan, generate, execute, and iterate.
- Language support and model selection for agent-driven tasks.
- Real-world examples ranging from small functions to multi-file features.
Enabling Agent Mode in Your IDE
- Activation procedures for VS Code, JetBrains, and Neovim.
- Configuring context window settings and model tier preferences.
- Establishing workspace rules and excluding large binary files.
- Managing the distinction between Copilot Chat and inline agent workflows.
Multi-Step Planning and Execution
- Prompting Copilot to develop features from start to finish.
- Observing how the agent decomposes tasks into steps across multiple files.
- Reviewing each step prior to applying changes.
- Utilizing inline rollback functions when steps deviate from the intended path.
Terminal Commands Within the Agent Loop
- Installing dependencies via Copilot's terminal integration.
- Executing build commands and interpreting their output.
- Managing environment variables directly from within Copilot sessions.
- Safety boundaries: identifying commands that require manual approval.
Test-Driven Development with an Agent
- Generating unit tests from existing source code.
- Facilitating test creation using natural language prompts.
- Running test suites and analyzing failure logs within Copilot.
- Refining assertions after encountering edge-case failures.
Navigating Large Codebases
- Automatically discovering cross-file references.
- Refactoring shared utilities with Copilot-guided renaming.
- Simultaneously updating configuration and schema files.
- Avoiding context window exhaustion through targeted prompts.
Customizing Copilot for Team Standards
- Writing repository-specific instructions in .github/copilot-instructions.md.
- Enforcing naming conventions and architectural patterns.
- Excluding sensitive files and directories from the context.
- Developing team-specific prompt templates for routine tasks.
GitHub Copilot Enterprise Governance
- Managing seat allocation, billing, and usage dashboards.
- Audit logs: tracking what Copilot generated versus what was committed.
- Microsoft IP indemnity policies and their licensing implications.
- Blocking specific file patterns from AI suggestion pipelines.
Debugging with Agent Mode
- Analyzing stack traces collaboratively with the agent.
- Hypothesis-driven debugging: asking Copilot why a test failed.
- 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 the appropriate models for different tasks.
- Monitoring agent latency and implementing caching strategies.
Security and Compliance for Enterprises
- Data handling protocols: determining what leaves your repository and 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 detect injection vulnerabilities.
Troubleshooting Common Scenarios
- Reasons why Copilot may ignore your codebase context.
- Resolving indexing failures for large repositories.
- Handling rate limit errors during peak usage hours.
- Fixing synchronization issues with IDE extensions.
Summary and Future Roadmap
- A recap of Agent Mode capabilities and practical workflows.
- GitHub's Copilot roadmap and upcoming agent features.
- Resources for staying updated on 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.
Target Audience
- Developers currently using Copilot who wish to unlock its agent mode capabilities.
- Engineering managers overseeing the rollout of Copilot across development teams.
- Security teams evaluating policies for AI-assisted code generation.
21 Hours
