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
- Distinguishing Copilot suggestions from agentic multi-step planning.
- The architecture of the agent loop: plan, generate, execute, iterate.
- Language support and model selection for agent tasks.
- Real-world examples: evolving 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 ignoring large binary files.
- Differentiating between Copilot Chat and inline agent workflows.
Multi-Step Planning and Execution
- Prompting Copilot to build a feature end-to-end.
- Observing the agent break tasks into steps across files.
- Reviewing each step before applying changes.
- Utilizing inline rollback when steps diverge from expectations.
Terminal Commands Inside the Agent Loop
- Installing dependencies through Copilot's terminal integration.
- Running build commands and interpreting output.
- Managing environment variables directly within Copilot sessions.
- Safety boundaries: identifying commands that require manual approval.
Test-Driven Development with an Agent
- Generating unit tests from existing source code.
- Driving test creation using natural language prompts.
- Running test suites and interpreting failure logs inside Copilot.
- Refining assertions after observing edge-case failures.
Navigating Large Codebases
- Automatically locating cross-file references.
- Refactoring shared utilities with Copilot-guided renames.
- Updating configuration and schema files in tandem.
- 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.
- Creating team-specific prompt templates for common 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 licensing implications.
- Blocking specific file patterns from AI suggestion pipelines.
Debugging with Agent Mode
- Analyzing stack traces together 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.
- Switching between models for different tasks.
- Monitoring agent latency and caching strategies.
Security and Compliance for Enterprises
- Data handling: understanding what leaves your repository and what stays local.
- Preventing the leakage of secrets and credentials via prompts.
- Ensuring compliance with GDPR, SOC 2, and FedRAMP requirements.
- Red-teaming generated code for injection vulnerabilities.
Troubleshooting Common Scenarios
- Understanding why Copilot may occasionally ignore your codebase context.
- Resolving indexing failures in large repositories.
- Handling rate limit errors during peak hours.
- Fixing IDE extension synchronization issues.
Summary and Future Roadmap
- Recap of Agent Mode capabilities and practical workflows.
- GitHub's Copilot roadmap and upcoming agent features.
- Resources for staying current with Copilot releases.
Requirements
- Experience with object-oriented or functional programming.
- A GitHub account and foundational knowledge of Git workflows.
- Familiarity with at least one IDE (VS Code, JetBrains, or Neovim).
Audience
- Developers currently using Copilot who wish to unlock agent mode capabilities.
- Engineering managers implementing Copilot across development teams.
- Security teams evaluating policies for AI-assisted code generation.
21 Hours
