Github Copilot CLI: Microsoft's UNLIMITED FREE Claude 4 Sonnet based CODER! Claude Code Alternative!

GitHub Copilot CLI Public Preview: A Detailed Overview

Key Concepts:

• GitHub Copilot CLI: A command-line interface agent powered by the same architecture as GitHub Copilot in VS Code.
• Public Preview: A phase where the software is available for public testing and feedback.
• Claude Code and Codeex: Other AI coding assistants.
• MCP (Memory Context Protocol) Server: A server that provides context and memory to the AI agent.
• Byte Rover: A memory layer that allows the AI coder to create and access memories.
• Sonnet: The default AI model used by Copilot CLI during the public preview.
• GPT-5: An alternative AI model that can be used by setting the copilot model environment variable.
• Read Bash and Write Bash: Tools that allow the AI agent to interact with interactive shell commands.

Introduction to GitHub Copilot CLI

GitHub has launched Copilot CLI in public preview, aiming to bring the power of its AI coding agent to the terminal. This tool competes with other AI coding assistants like Claude Code and Codeex. It's based on the same Copilot agent architecture that powers Copilot in VS Code and GitHub. The key advantage is the ability to code, script, and interact with GitHub repositories, issues, and pull requests directly from the command line using natural language.

Key Features and Functionality

• Direct Command-Line Access: Allows users to work with the Copilot coding agent directly in the command line, eliminating context switching.
• GitHub Integration: Provides access to repositories, issues, and pull requests using natural language, authenticated with the user's existing GitHub account.
• Code Collaboration: Enables building, editing, debugging, and refactoring code with an AI collaborator that can plan and execute complex tasks.
• MCP Server Support: Ships with GitHub's MCP server by default and supports custom MCP servers to extend capabilities.
• Preview and Approval: Requires explicit approval before executing any action, ensuring user control.
• Easy Installation: Can be installed with a single command.
• Model Flexibility: Uses Claude Sonnet by default but can be switched to GPT-5 via the copilot model environment variable.

Access and Pricing

The documentation states that GitHub Copilot is available with GitHub Copilot Pro, GitHub Copilot Pro Plus, GitHub Copilot Business, and GitHub Copilot Enterprise plans. However, during the public preview, it is working without any rate limits or restrictions, even without a paid plan. This allows users to use it for free with Sonnet.

Installation and Setup

The installation process involves a single command. Once installed, running the copilot command starts the CLI. If the user already has the Git CLI, GitHub Desktop, or Copilot in VS Code set up, it authenticates from there.

Slash Commands and MCP Servers

The CLI provides slash commands like /add directory, which allows users to limit file access to specific directories. The /mcp command allows adding MCP servers. The presenter recommends adding Byte Rover, a memory layer that allows the AI coder to create memories of important information. Byte Rover memories can be synced across teams and updated like Git.

Unique Tools: Read Bash and Write Bash

Copilot CLI features unique tools called read bash and write bash, which allow it to interact with interactive shell commands. This enables it to handle commands that require user input in the terminal.

Testing and Performance

The presenter tested Copilot CLI with several tasks:

1. Movie Tracker App (Expo): The CLI successfully built a movie tracker app using Expo. The output was comparable to Claude Code, with minor issues like missing padding or margin. The presenter noted that the Copilot CLI's output was slightly better than Claude Code's.
2. Go Terminal: The CLI successfully built a functional Go terminal on the first try, which is a task that Claude Code often struggles with.
3. FPS Go Game Modification: The CLI successfully added a step counter and a health bar to an existing FPS Go game. However, the implementation took longer than with DeepSseek or Claude Code, and the CLI exhibited "overesting" behavior, repeatedly rechecking and attempting to fix unrelated issues.
4. SVG Generation Model (Open Code Repo): This task failed.

Overall Assessment

The presenter ranks Copilot CLI in third place on their leaderboard of AI coding assistants. The key advantages are its current free availability, the use of Sonnet without rate limits, and the unique read bash and write bash tools. The main drawback is the "overesting" behavior observed during the FPS Go game modification task.

Conclusion

GitHub Copilot CLI is a promising AI coding assistant that offers a unique set of features and capabilities. Its current free availability during the public preview makes it a worthwhile tool to try. The presenter encourages viewers to share their thoughts and subscribe to the channel.