Allison reports on GitHub Copilot CLI’s latest weekly updates, highlighting multiline input, server setup improvements, enhanced noninteractive workflows, and support for Anthropic Claude Haiku 4.5.

GitHub Copilot CLI: Multiline Input, MCP Enhancements, and Haiku 4.5 Release

Author: Allison

Stay updated on the ongoing improvements to GitHub Copilot CLI, directly influenced by developer feedback.

Highlights This Week

Multiline Input Support

  • Copilot CLI now allows multiline command input using Shift+Enter in terminals supporting the Kitty protocol.
  • In Visual Studio Code (VSCode) and compatible forks, enable multiline input via /terminal-setup.

Anthropic Claude Haiku 4.5 Model Support

  • The CLI integrates with the new Anthropic Claude Haiku 4.5 AI model, with a playful changelog note marking its debut.

MCP (Multi-Command Proxy) Server Enhancements

  • Easier server setup: The command field now accepts full shell commands, replacing the older comma-separated value format.
  • Environment variables can now be entered using common shell syntax (${VARIABLE_NAME}), improving the developer experience and consistency with other CLI tools.
  • Bug fixes boost reliability for tool calls in MCP sessions.

Expanded Noninteractive Workflow Capabilities

  • PAT and gh authentication now respect the GH_HOST environment variable, making it possible for GitHub Enterprise Cloud customers with data residency requirements to use noninteractive authentication seamlessly.
  • Permissions requests are better handled, and a new --allow-all-paths flag is available for granting broad permissions in noninteractive mode.

Stability Improvements and Forward-Looking Changes

  • Session storage and management have been refactored for clarity, conciseness, and scalability.
  • The CLI now cleanly separates data storage logic from the display, setting the stage for future integrations and features.
  • Model premium request multipliers are now more visible.
  • PowerShell support is significantly improved and no longer considered experimental.
  • Debugging and logging are enhanced: Stack traces, Copilot API request IDs, and additional debug information are easier to extract. The log level can be set persistently in the config file.

How to Update and Provide Feedback

  • Update Copilot CLI by running:

    npm install -g @github/copilot@latest

  • Contribute with feedback using the /feedback command or by opening issues in the GitHub Copilot CLI public repository.

Developer input continues to shape Copilot CLI, bringing quality-of-life upgrades and new features that make command-line development smoother.

This post appeared first on “The GitHub Blog”. Read the entire article here