Demo: Using GitHub Copilot CLI and yolo mode

Key Concepts

• GitHub Copilot CLI: A command-line interface for interacting with GitHub Copilot.
• Linting: The process of analyzing code for potential errors, stylistic issues, and adherence to coding standards.
• YOLO Mode (d-yolo): A setting within Copilot CLI that grants the tool unrestricted access to tools, paths, and URLs, enabling automated execution without prompting for approval.
• Tool Calling: Copilot’s ability to utilize external tools (like linters) to analyze and improve code.
• MCPS: (Mentioned briefly, likely referring to Machine Code Protection Services or a similar security feature – context is limited).

Introduction to GitHub Copilot CLI and Code Improvement

Scott demonstrates the functionality of the GitHub Copilot Command Line Interface (CLI) focusing on its ability to analyze and improve a TypeScript codebase. He begins by posing a question to Copilot: “Are there any improvements that we can make to this codebase?” specifically noting his unfamiliarity with TypeScript and a desire to leverage linting for code quality. He initially uses Copilot with standard permissions, which require explicit approval for tool usage. He previously used a tool called Handy (handy.computer) for dictation, highlighting its speed advantage.

The "YOLO Mode" Functionality

The core demonstration revolves around Copilot’s “YOLO mode” (activated with the d-yolo flag). This mode bypasses the approval prompts for tool access, path access, and URL access, allowing Copilot to autonomously execute tasks. Scott emphasizes the potential risks associated with YOLO mode, suggesting running it within a trusted environment or a Docker container. He explains that YOLO mode equates to allowing all tools, paths, and URLs.

He illustrates this by re-running the initial request – to lint the TypeScript application – after enabling YOLO mode. This time, Copilot directly invokes the linter without requesting permission, showcasing the automated workflow. Scott references “Ralph Wiggams loops” as a potential outcome of YOLO mode, implying a rapid, iterative process of automated code analysis and improvement.

Permissions and Control Options

Scott highlights the available permission controls within Copilot CLI, accessible via copilot help. These options include:

• Allow all paths: Grants Copilot access to all file system paths.
• Allow tools: Permits Copilot to utilize external tools.
• Deny tools: Restricts Copilot from using external tools.
• Allow URLs: Enables Copilot to access external URLs.

He reiterates that YOLO mode effectively activates all these permissions simultaneously.

Installation and Resources

Scott provides installation instructions for Copilot CLI, noting its availability through:

• winget install Copilot (for Windows)
• brew (for macOS)
• npm copilot (for Node Package Manager)
• copilot.github.com (for general information and downloads)
• github.com (GitHub’s main website)

Synthesis

The demonstration showcases GitHub Copilot CLI’s powerful capabilities for automated code analysis and improvement, particularly through the YOLO mode. While YOLO mode offers significant efficiency gains, Scott stresses the importance of exercising caution and utilizing it within secure environments. The video provides a practical overview of Copilot CLI’s functionality, permission controls, and installation methods, offering actionable insights for developers seeking to enhance their coding workflows.