What’s new in the GitHub Copilot CLI?

Key Concepts

• Copilot SDK: Software Development Kit for building applications leveraging GitHub Copilot’s capabilities. Available in Node.js/TypeScript, Python, Go, and .NET.
• Multi-turn Conversations: The ability for applications to engage in extended, contextual dialogues with users.
• Tool Execution: Copilot’s capacity to utilize external tools and APIs as part of its workflow.
• GitHub Copilot CLI: Command Line Interface for interacting with GitHub Copilot, enhanced with new features and models.
• Custom Agents: Pre-built agents within the Copilot CLI designed for specific tasks (explore, task planning, code review).
• Agent Skills: Capabilities that can be added to custom agents to enhance their functionality.

Copilot SDK – Technical Preview & Capabilities

The GitHub Copilot SDK is currently in technical preview, offering developers the ability to integrate Copilot’s intelligence into their own applications. The SDK is available across multiple programming languages: Node.js/TypeScript, Python, Go, and .NET. This allows for the creation of applications capable of handling multi-turn conversations, meaning the application can maintain context and engage in extended dialogues with users. Crucially, the SDK also supports tool execution, enabling Copilot to interact with and utilize external tools and APIs – expanding its functionality beyond code generation. Finally, the SDK provides full life cycle control, giving developers granular control over the interaction process. Early adopters are already building applications utilizing these features.

GitHub Copilot CLI – Recent Updates & Installation

Significant updates have been rolled out to the GitHub Copilot CLI in recent weeks. These include access to a wider range of models, enhancing the versatility of the tool. Furthermore, several built-in custom agents have been added, specifically designed for tasks such as:

• Explore: Likely refers to exploratory code analysis or discovery.
• Task Plan: Automated planning of development tasks.
• Code Review: Automated assistance with code review processes.

These custom agents can be further augmented with agent skills, creating a robust and adaptable workflow. Installation of the Copilot CLI is now simplified through package managers like Homebrew and Windget, alongside dedicated install scripts. A key benefit of these installation methods is automatic updates, ensuring users always have the latest version. However, standalone releases remain available for users with specific distribution requirements.

CLI Enhancements & Scripting

Beyond the core features, numerous improvements have been made to the Copilot CLI. These include new flags for scripting, allowing for more automated and integrated workflows. Improvements to the diff view (likely referring to the display of code changes) and the addition of web controls further enhance the user experience. The breadth of these updates suggests a continuous development cycle focused on usability and functionality.

Feedback & Community Engagement

The development team is actively soliciting user feedback to guide future development. Feedback can be provided through multiple channels:

• Comments on the YouTube video hosting this announcement.
• Social media platforms.
• Creating issues or participating in discussions within the Copilot repository.

As stated, the team is “actively seeking out your feedback” and welcomes “wish list items” and general suggestions.

Synthesis

The recent advancements in both the Copilot SDK and CLI demonstrate a significant push towards making GitHub Copilot a more versatile and integrated development tool. The SDK empowers developers to build custom applications leveraging Copilot’s AI capabilities, while the CLI enhancements streamline workflows and provide more powerful features for direct interaction. The emphasis on community feedback highlights a commitment to iterative improvement and user-driven development. The combination of these updates positions Copilot as a rapidly evolving platform with substantial potential for enhancing developer productivity.