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: Grasping the Shift
- Distinguishing Copilot suggestions from agentic multi-step planning.
- Understanding the agent loop architecture: plan, generate, execute, and iterate.
- Exploring language support and model selection for agent tasks.
- Reviewing real-world examples, ranging from five-line functions to multi-file features.
Enabling Agent Mode in Your IDE
- Activation procedures for VS Code, JetBrains, and Neovim.
- Configuring context windows and model tier preferences.
- Setting 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 build features from start to finish.
- Observing the agent break down tasks into 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 from Copilot sessions.
- Understanding safety boundaries and identifying commands that require manual approval.
Test-Driven Development with an Agent
- Generating unit tests based on existing source code.
- Driving test creation using natural language prompts.
- Running test suites and interpreting failure logs within Copilot.
- Refining assertions after identifying edge-case failures.
Navigating Large Codebases
- Automatically locating cross-file references.
- Refactoring shared utilities with Copilot-guided renaming.
- Updating configuration and schema files simultaneously.
- Preventing 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 context analysis.
- Creating team-specific prompt templates for common tasks.
GitHub Copilot Enterprise Governance
- Managing seat allocation, billing, and usage dashboards.
- Utilizing audit logs to track what Copilot generated versus what was committed.
- Blocking specific file patterns from AI suggestion pipelines.
Debugging with Agent Mode
- Analyzing stack traces in collaboration with the agent.
- Using agent-assisted bisect to identify regression sources.
- Managing hallucination risks when debugging unfamiliar code.
Performance and Limit Management
- Understanding daily request limits and model quotas.
- Optimizing prompt length to avoid truncated responses.
- Switching between models for different tasks.
- Monitoring agent latency and caching strategies.
Security and Compliance for Enterprises
- Understanding data handling: what leaves your repository and what stays local.
- Preventing leakage of secrets and credentials via prompts.
- Compliance with GDPR, SOC 2, and FedRAMP requirements.
- Red-teaming generated code for injection vulnerabilities.
Troubleshooting Common Scenarios
- Understanding why Copilot sometimes ignores your codebase context.
- Resolving indexing failures for large repositories.
- Handling rate limit errors during peak hours.
- Fixing IDE extension sync issues.
Summary and Future Roadmap
- Recap of Agent Mode capabilities and practical workflows.
- Resources for staying current with 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 already using Copilot who wish to unlock and master agent mode.
- Engineering managers overseeing the rollout of Copilot across development teams.
- Security teams responsible for reviewing policies on AI-assisted code generation.
21 Hours
