GitHub Copilot CLI Beginner's Guide: Getting Started with GitHub Copilot CLI

GitHub Copilot CLI is an agentic AI tool that lets you give coding instructions inside the terminal, and hand off tasks like file exploration, code generation, and running tests. It leverages repository context as-is, letting you continue your work within a console-based workflow.

Installation is simple via npm.

• npm install -g @github/copilot
• If you have Node, this is the most basic installation path
• You can also install it via WinGet or Homebrew

After installation, run Copilot from the command line, and on first use, authenticate to your GitHub account with /login. This process links your Copilot account, and also connects the read-only GitHub MCP server.

You also need to approve folder access permissions. You can allow it for a single session only, or save it so it applies to future sessions as well, letting you reopen the same project without repeated approvals.

Real-world usage examples are just as simple. Ask about a project overview, and it explores the files and summarizes them; give it a code task like adding a new endpoint, and it references existing docs and examples to try to follow the project's conventions. Well-defined tasks can be handed off to the Copilot cloud agent with /delegate, in which case it retains the context of the current session, creates a new branch, opens a draft pull request, and proceeds with changes in the background.

The next installment covers two execution modes. The key ones are interactive mode, for handling projects directly, and non-interactive mode, which uses the -p flag to quickly get summaries without leaving the shell context.