关于生成 Copilot 扩展

About Copilot Extensions

Copilot Extensions are integrations that expand the functionality of Copilot Chat, allowing developers to bring external tools, services, and custom behaviors into the Chat experience. You can use Copilot Extensions to extend the capabilities of Copilot Chat in a variety of ways, including:

• Querying documentation: A Copilot Extension can allow Copilot Chat to query a third-party documentation service to find information about a specific topic.
• AI-assisted coding: A Copilot Extension can use a third-party AI model to provide code suggestions.
• Data retrieval: A Copilot Extension can allow Copilot Chat to query a third-party data service to retrieve information about a specific topic.
• Action execution: A Copilot Extension can allow Copilot Chat to execute a specific action, such as posting to a message board or updating a tracking item in an external system.

About building GitHub Copilot Extensions

GitHub Copilot Extensions are a type of Copilot Extension built with GitHub Apps. GitHub Copilot Extensions are best suited for developers who want cross-platform compatibility and app management and support from GitHub.

Supported clients and IDEs

Clients and IDEs	GitHub Copilot Extensions support
Visual Studio Code
Visual Studio
GitHub.com
GitHub Mobile
JetBrains IDEs
GitHub Codespaces
Vim/Neovim
Copilot in the CLI
Xcode

About visibility of GitHub Copilot Extensions

GitHub Copilot Extensions can be private, public and shareable, or public and listed on the GitHub Marketplace. Which visibility option you choose will depend on your use case and the audience you are targeting.

• Private extensions are often preferred by large enterprises or companies that:
  • Want more customization and controls over data access
  • Need to integrate with a large volume of internal documents and databases
  • Have strict security policies making it difficult to authorize permissions for third-parties
• Public extensions are suitable for:
  • Open-source projects
  • Collaborative development and use across organizations within an enterprise
  • Sharing your tool and getting feedback before publishing to the GitHub Marketplace
• GitHub Marketplace extensions are ideal for third-parties that want to:
  • Offer their service to a broader audience
  • Integrate their tool into the developer workflow on GitHub and the IDE
  • Leverage the GitHub ecosystem to raise awareness for their product

About skillsets and agents

Skillsets and agents are the two ways to extend Copilot's capabilities and context through the Copilot Extensibility Platform. They let you integrate external services and APIs into Copilot Chat, but each one serves different use cases and offers different levels of control and complexity:

• Skillsets are lightweight and streamlined, designed for developers who need Copilot to perform specific tasks (e.g., data retrieval or simple operations) with minimal setup. They handle routing, prompt crafting, function evaluation, and response generation automatically, making them ideal for quick and straightforward integrations. For more information about skillsets, see About Copilot skillsets.
• Agents are for complex integrations that need full control over how requests are processed and responses are generated. They let you implement custom logic, integrate with other LLMs and/or the Copilot API, manage conversation context, and handle all aspects of the user interaction. While Agents require more engineering and maintenance, they offer maximum flexibility for sophisticated workflows. For more information about agents, see "About Copilot agents."

About context passing

You can allow your Copilot Extension to receive context from the editor, such as the currently opened file, by enabling the Read-only access level for the "Copilot Editor Context" permission in your GitHub App settings. See Creating a GitHub App for your Copilot Extension.

The GitHub Copilot Extensibility Platform automatically handles messaging when implicit and explicit context is unavailable or unauthorized. To enable context passing, you are required to request permissions from users. When requesting permissions, follow these best practices:

• Clearly communicate what context you need and what you need it for.
• Implement appropriate error handling for unavailable context that your own application logic and API calls.
• In the event context is unavailable, provide value where possible without this data.
• Request only the minimum required permissions for your extension.

Context passing respects content exclusions, .env files, and files listed in the content exclusion settings.

Resources for building GitHub Copilot Extensions

GitHub provides a comprehensive toolkit for extension builders, with code samples, a CLI debugging tool, quickstart SDKs, and a user feedback repository. For more information, see the copilot-extensions organization on GitHub.

Before creating your own GitHub Copilot Extension from scratch, you may want to explore an existing Copilot agent, then integrate it with a GitHub App to see how it works. GitHub provides a few example Copilot agents that you can clone and use as the basis for your own GitHub Copilot Extension:

• Blackbeard: A simple Copilot agent that responds to requests like a pirate, using Copilot's LLM API and special system prompts. It is a good starting point for learning how to build a GitHub Copilot Extension. For more information, see the Blackbeard Copilot Extension.

• GitHub Models: A more complex Copilot agent that lets you ask about and interact with various LLMs listed on the GitHub Marketplace from within Copilot Chat. For more information, see the GitHub Models Copilot Extension.

  Note

  GitHub Models are in public preview and subject to change. To request access, join the waitlist.

• Function calling: an example agent written in Go that demonstrates function calling and confirmation dialogues. For more information, see the Function calling extension.

• RAG extension: an example agent written in Go that demonstrates a simple implementation of retrieval augmented generation. For more information, see the RAG extension.

• Preview SDK: An SDK that simplifies the process of building GitHub Copilot Extensions by handling request verification, response formatting, and API interactions. It allows builders to focus on their extension's core functionality rather than boilerplate, by streamlining the integration of tools, APIs, and data sources into Copilot Chat. For more information, see the Preview SDK.

About building Copilot-enabled Visual Studio Code extensions

You can build a Copilot Extension that is exclusive and native to Visual Studio Code, called a Copilot-enabled Visual Studio Code extensions. This option is best suited for developers who want to build extensions that use VS Code-specific APIs and functionality, or extend existing VS Code extensions.

Also known as VS Code Chat extensions, Copilot-enabled Visual Studio Code extensions function similarly to GitHub Copilot Extensions by extending the capabilities of Copilot Chat, with a few notable differences:

• VS Code Chat extensions are only usable in VS Code.
• VS Code Chat extensions have more access to VS Code's features and APIs, allowing more editor-specific interactions like accessing local workspace data, manipulating Visual Studio Code's interface, and read/write access to local files.
• VS Code Chat extensions are published to the VS Code Marketplace, not the GitHub Marketplace.
• VS Code Chat extensions are local to the user's machine, and cannot be controlled by an organization or enterprise's policies.

For more information on Copilot-enabled Visual Studio Code extensions, see Chat extensions in the Visual Studio Code documentation.

Further reading

• Copilot Extensions Glossary