在 IDE 中向 GitHub Copilot 提问

Prerequisites

• Access to GitHub Copilot. See "What is GitHub Copilot?."
• Latest version of Visual Studio Code. See the Visual Studio Code download page.
• The GitHub Copilot extension - Install this from the Visual Studio Marketplace. For more information, see "Set up GitHub Copilot in Visual Studio Code" in the Microsoft documentation.
• Sign in to GitHub in Visual Studio Code. If you experience authentication issues, see "Troubleshooting issues with GitHub Copilot Chat in IDEs."

If you have access to GitHub Copilot via your organization, you won't be able to use GitHub Copilot Chat if your organization owner has disabled chat. See "Managing policies for Copilot in your organization."

Submitting prompts

You can ask Copilot Chat to give code suggestions, explain code, generate unit tests, and suggest code fixes.

1. To open the chat view, click the chat icon in the activity bar or press Control+Command+i (Mac) / Ctrl+Alt+i (Windows/Linux).

   

   Tip

   For additional ways to access Copilot Chat, including inline with your code, see Additional ways to access Copilot Chat below.

2. Enter a prompt in the prompt box, or click one of the suggested prompts. For example prompts, see "Example prompts for Copilot Chat."

3. Evaluate Copilot's response, and make a follow up request if needed.

   The response may contain text, code blocks, buttons, images, URIs, and file trees. The response often includes interactive elements. For example, the response may include a menu to insert a code block, or a button to invoke a Visual Studio Code command.

   To see the files that Copilot Chat used to generate the response, select the Used n references dropdown at the top of the response. The references may include a link to a custom instructions file. This file contains additional information that is automatically added to all of your chat questions to improve the quality of the responses. For more information, see "Adding custom instructions for GitHub Copilot."

Using keywords in your prompt

You can use special keywords to help Copilot understand your prompt. For examples, see "Example prompts for Copilot Chat."

Chat participants

Chat participants are AI domain experts that can perform tasks or answer questions in a specific domain.

Copilot Chat can infer relevant chat participants based on your natural language prompt, improving discovery of advanced capabilities without you having to explicitly specify the participant you want to use in your prompt. However, if you prefer, you can still manually specify a chat participant to scope your prompt to a specific domain. To do this, type @ in the chat prompt box, followed by a chat participant name.

Chat participants include:

• @workspace: Has context about the code in your workspace. Use @workspace when you want Copilot to consider the structure of your project, how different parts of your code interact, or design patterns in your project.
• @vscode: Has context about Visual Studio Code commands and features. Use @vscode when you want help with Visual Studio Code.
• @terminal: Has context about the Visual Studio Code terminal shell and its contents. Use @terminal when you want help creating or debugging terminal commands.
• @azure: Has context about Azure services and how to use, deploy and manage them. Use @azure when you want help with Azure. The @azure chat participant is currently in public preview and is subject to change.

In addition to the built-in Visual Studio Code chat participants, you can also install Copilot Extensions that provide chat participants. You can install these extensions from GitHub Marketplace and from Visual Studio Code Marketplace. For information about extensions from GitHub Marketplace that provide chat participants, see "Using extensions to integrate external tools with Copilot Chat."

To see all available chat participants, type @ in the chat prompt box.

See also Chat participants in the Visual Studio Code documentation.

Slash commands

Use slash commands to avoid writing complex prompts for common scenarios. To use a slash command, type / in the chat prompt box, followed by a command. Slash commands include:

• /tests: Generate unit tests for the selected code
• /fix: Propose a fix for problems in the selected code
• /explain: Explain the selected code
• /clear: Start a new chat

To see all available slash commands, type / in the chat prompt box. See also Slash commands in the Visual Studio Code documentation.

Chat variables

Use chat variables to include specific context in your prompt. To use a chat variable, type # in the chat prompt box, followed by a chat variable. Chat variables include:

• #file: Include a specific file as context in the chat.
• #git: Include information about the current Git repository.
• #terminalLastCommand: Include the last run command in the active Visual Studio Code terminal.

To see all available chat variables, type # in the chat prompt box. See also Chat variables in the Visual Studio Code documentation.

Using GitHub skills for Copilot

Copilot's GitHub-specific skills expand the type of information Copilot can provide. To access these skills in Copilot Chat in VS Code, include @github in your question.

When you add @github to a question, Copilot dynamically selects an appropriate skill, based on the content of your question. You can also explicitly ask Copilot Chat to use a particular skill. You can do this in two ways:

• Use natural language to ask Copilot Chat to use a skill. For example, @github Search the web to find the latest GPT4 model from OpenAI.
• To specifically invoke a web search you can include the #web variable in your question. For example, @github #web What is the latest LTS of Node.js?

Currently available skills

You can generate a list of currently available skills by asking Copilot: @github What skills are available?

The skills you can use in Copilot Chat in VS Code include those shown in the table below.

Skill	Description	Enabled by default?	Example question
Commit details	Retrieves a list of commits, or the contents of a specific commit, to provide answers to commit-related questions.	Yes	@github Explain the changes in the code of this commit
Discussion details	Retrieves a specific GitHub discussion. This is useful for quickly getting the gist of the conversation in a discussion.	Yes	@github Summarize this discussion LINK-TO-DISCUSSION
File details	Retrieves a specific file in the default branch of the Git repository, allowing you to ask questions about the file and the recent changes made to it. This skill is useful when you provide the exact path of a file in the repository.	Yes	@github What logic does user_auth.js encapsulate? @github What is the file history of user_auth.js?
GitHub Advanced Security	Retrieves information about security alerts within your organization from GitHub Advanced Security features (code scanning, secret scanning, and Dependabot alerts).	Yes	@github How would I fix this code scanning alert?
Issue details	Retrieves a specific GitHub issue, including the issue's title, number, author, status, body, linked pull requests, comments, and timestamps.	Yes	@github List issues assigned to me
Lexical code search	Keyword code search in the default branch of the Git repository. This skill is useful when you want to know about specific functions, methods or keywords that exist in the code. This skill leverages most of the functionality available to GitHub Search like symbol and path.	Yes	Find me the tests for the GitService class
Path search	Retrieves a specific file in the default branch of the Git repository. This skill is useful when you provide the exact path of a file in the repository.	Yes	@github What logic does user_auth.js encapsulate?
Pull request details	Retrieves a specific pull request. This allows you to ask questions about the pull request, including getting a summary of the pull request, its comments, or the code it changes.	Yes	@github List my recent pull requests
Release details	Retrieves the latest, or specified, release. This allows you to find out who created a release, when it happened, and information included in the release notes.	Yes	@github When was the latest release?
Repository details	Retrieves a specific GitHub repository. This is useful for finding out details such as the repository owner and the main language used.	Yes	@github Tell me about PATH-TO-REPOSITORY
Semantic code search	Natural language semantic code search in the default branch of the Git repository. This skill is useful when you want to know where or how certain functionality has been implemented in the code. Note: this requires indexing to be enabled for the repository (see "Indexing repositories for Copilot Chat").	Yes Available for all repositories with a Copilot Enterprise subscription, and for public repositories with a Copilot Individual or Copilot Business subscription.	How does this repo manage HTTP requests and responses?
Support search	Retrieves information from the GitHub Support portal. This skill is useful for asking Copilot Chat about GitHub products and support related questions.	Yes	@github Can I use Copilot knowledge bases with Copilot Individual?
Web search	Searches the web using the Bing search engine. This skill is useful for teaching Copilot about recent events, new developments, trends, technologies, or extremely specific, detailed, or niche subjects.	No For Copilot Individual: Enable in your user settings. For Copilot Business: Enable in organization settings.	@github #web What are some recent articles about SAT tokens securing against vulnerabilities in Node?

AI models for Copilot Chat

The following models are currently available through multi-model Copilot Chat:

• GPT 4o: This is the default Copilot Chat model. It is a versatile, multimodal model that excels in both text and image processing and is designed to provide fast, reliable responses. It also has superior performance in non-English languages. Learn more about the model's capabilities and review the model card. Gpt-4o is hosted on Azure.
• Claude 3.5 Sonnet: This model excels at coding tasks across the entire software development lifecycle, from initial design to bug fixes, maintenance to optimizations. Learn more about the model's capabilities or read the model card. GitHub Copilot uses Claude 3.5 Sonnet hosted on Amazon Web Services.
• o1-preview: This model is focused on advanced reasoning and solving complex problems, in particular in math and science. It responds more slowly than the gpt-4o model. You can make 10 requests to this model per day. Learn more about the model's capabilities and review the model card. o1-preview is hosted on Azure.
• o1-mini: This is the faster version of the o1-preview model, balancing the use of complex reasoning with the need for faster responses. It is best suited for code generation and small context operations. You can make 50 requests to this model per day. Learn more about the model's capabilities and review the model card. o1-mini is hosted on Azure.

For more information about the o1 models, see Models in the OpenAI Platform documentation.

For more information about the Claude 3.5 Sonnet model from Anthropic, see "Using Claude 3.5 Sonnet in GitHub Copilot."

Changing your AI model

1. If you access Copilot Chat through a Copilot Business subscription, your organization must grant members the ability to switch to a different model. See "Managing policies for Copilot in your organization."
2. To open the chat view, click the chat icon in the activity bar or press Control+Command+i (Mac) / Ctrl+Alt+i (Windows/Linux).
3. In the bottom right of the chat view, select the CURRENT-MODEL dropdown menu, then click the AI model of your choice.

Additional ways to access Copilot Chat

In addition to submitting prompts through the chat view, you can submit prompts in other ways:

• Inline: To start an inline chat directly in the editor or integrated terminal, enter Command+i (Mac) / Ctrl+i (Windows/Linux).
• Quick chat: To open the quick chat dropdown, enter Shift+Command+i (Mac) / Shift+Ctrl+i (Windows/Linux)
• Smart actions: To submit prompts via the context menu, right click in your editor, select Copilot in the menu that appears, then select one of the actions. Smart actions can also be accessed via the sparkle icon that sometimes appears when you select a line of code.

See inline chat, quick chat, and chat smart actions in the Visual Studio Code documentation for more details.

Multi-file edits

Use Copilot Edits to make changes across multiple files directly from a single Copilot Chat prompt.

1. Enable Copilot Edits in the Copilot Extension settings.
2. Start an edit session by selecting Open Copilot Edits from the Copilot Chat menu.
3. Optionally, add relevant files to the working set to indicate to GitHub Copilot which files you want to work on.
4. Submit a prompt. In response to your prompt, Copilot Edits determines which files in your working set to change and adds a short description of the change.
5. Review the changes and Apply or Discard the edits for each file.

For more information, see Copilot Edits in the Visual Studio Code documentation.

Sharing feedback

To indicate whether a response was helpful, use the thumbs up and thumbs down icons that appear next to the response.

To leave feedback about the GitHub Copilot Chat extension, open an issue in the microsoft/vscode-copilot-release repository.

Further reading

• "Prompt engineering for GitHub Copilot"
• Using Copilot Chat in VS Code and Getting started with GitHub Copilot Chat in VS Code in the Visual Studio Code documentation
• "Asking GitHub Copilot questions in GitHub"
• "Responsible use of GitHub Copilot Chat in your IDE"
• "GitHub Terms for Additional Products and Features"
• GitHub Copilot Trust Center
• GitHub Copilot FAQ