Microsoft-Cloud-Native

GitHub Copilot Cheat Sheet

This is a quick reference guide to get started with GitHub Copilot.
This page is a fork from: jaywcj

Preparation

1. Account Registration

You need to have a GitHub account and subscribe to Copilot.

Item	Description
GitHub Account	Registration Link
Subscribe to GitHub Copilot	Subscription Link

2. Install Vscode Extension

Search and install the following extensions in the extension marketplace.

Extension Name	Functionality
`GitHub Copilot`	Code completion suggestions in the editor
`GitHub Copilot Chat`	Plugin bar for conversing with Copilot

After installation, click on the GitHub Copilot plugin icon in the bottom right corner and click Sign in to GitHub to log in.
Alternatively, click on the Accounts icon in the toolbar and click Sign in with GitHub to use GitHub Copilot to log in.

4. Copilot Subscription Plans

Plan | Price | Features :-|-|- Copilot Individual | $10/person/month
(Free for students, teachers, and open-source contributors) | Code completion, Chatbot Copilot Business| $19/person/month | Code completion, Chatbot, Command-line tool, Security vulnerability screening, Code references, Public code screening, Intellectual property, Enterprise security and privacy protection Copilot Enterprise| $39/person/month | Business features + Personalized chat for private code repositories + Document search summaries + Git Pull Request summaries + Code review + Model fine-tuning

5. Copilot Usage Entry Points

Name	Description
Inline Suggestions	Show suggestions next to the cursor in the editor
Completions Panel	Display a complete list of suggestions in the editor
Inline Chat	Initiate a conversation next to the cursor in the editor
Editor Chat	Open a full conversation interface in the editor
Slide Chat	Open a conversation interface in the sidebar of the editor
Quick Chat	Invoke the conversation interface at the top

Tips for Suggestions

The Art of Suggestions

The relationship between you and copilot is like that of a writer and an illustrator.
You need to describe your story (i.e., context) as comprehensively, concisely, and clearly as possible.
Copilot will then generate beautiful illustrations (i.e., code) based on your story.

Provide context information
Context can be predicted

Practical Tutorials

Types of Context Information

Type|Description

|- File|Copilot will analyze the currently open file and adjacent files in the editor. Comments|Copilot will provide help and suggestions based on adjacent comments, such as docstrings, block comments, and line comments. Naming|Good naming helps Copilot better understand your code, such as function names, variable names, and file names. Code|Copilot will analyze your code and the surrounding code to generate helpful suggestions.

Context: File

Copilot will analyze the currently open file and adjacent files in the editor to provide appropriate suggestions based on the context.

Avoid opening too many files so that Copilot can better understand your code.

Open files that are relevant and have common characteristics.

For new projects, you can open some template code, data files, and reference documents as example files. This will help Copilot better understand your expectations. Once you have developed some code, you can delete these example files.

Context: Comment: Top-level comment

When creating a new file, add comments at the top of the file to describe your requirements. This is helpful for Copilot.

* The following explanations will use ... to indicate where Copilot starts generating.

# Download file from an URL and analyze its content
# Details: 
# * Download the file from an URL
# * Save the downloaded files into `./download` folder
# * Use `filetype` of the file to specify how to parse
# * Filetype can be `.pdf`, `.html`, `.epub`, `.md` and `.txt`
# * Use NLP or OCR to get the file content
# * Tokenize the file content and get the statistics result

import  ...

Context: Comment: Inline Comment

Add comments above each function or important code block to help Copilot understand some intentions or issues in your code.

* The following instructions will use ... to indicate where Copilot starts generating.

Add comments above the function

# parse the JSON string into User object
def ...

Add comments to the code

# ...
api_sever = FastApi(...)

# starting the API Sever, enable ssl, bind to 8443 port
...

Context: Comment: Docstring

Sometimes, when you already have a detailed design document but haven’t written the functional code, you can directly use the description in the docstring to let Copilot generate the code.

def send_email(to_address: Email, subject: str, content: HTML): -> StatusCode:
    """
    Send email to specified address

    Parameters
    ----------
    to_address : Email
        The email address to send to
    subject : str
        The email subject
    content : HTML
        The email content

    Returns
    -------
    StatusCode
        The sending result
    """

    ...

Context: Comment: Asking questions

If you don’t want to switch to Copilot chat, comments can also be used to ask questions.

# Q: What is the difference between `os.path.join` and `pathlib.PurePath`?
# A: ...

Context: Comment: Todo

You can also let Copilot generate a todo list for you to assess the workload.

# Parse the json file into a Talks object
# TODO:
# -[ ] 1. ...

Context: Naming

Your naming should be clear enough for Copilot to understand your intent.

bad case

a = 60

def send(dict):
    ...

class data:
    ...

good case

timeout = 60

def send_email(to_address: Email, subject: str, content: HTML): -> StatusCode:
    ...

class Email:
    ...

Context: Code: Code examples

Provide code snippets to help Copilot start new development tasks better.

Frameworks and libraries used

Code style

Algorithm logic

from typing import List
from typing import Optional
from sqlalchemy import ForeignKey
from sqlalchemy import String
from sqlalchemy.orm import DeclarativeBase
from sqlalchemy.orm import Mapped
from sqlalchemy.orm import mapped_column
from sqlalchemy.orm import relationship

class Base(DeclarativeBase):
    pass

class User(Base):
    __tablename__ = "user_account"
    id: Mapped[int] = mapped_column(primary_key=True)
    name: Mapped[str] = mapped_column(String(30))
    fullname: Mapped[Optional[str]]
    addresses: Mapped[List["Address"]] = relationship(
        back_populates="user", cascade="all, delete-orphan"
    )
    def __repr__(self) -> str:
        return f"User(id={self.id!r}, name={self.name!r}, fullname={self.fullname!r})"

# Email Address
...

Context: Code: Data examples

Provide data snippets to help Copilot start new development tasks better.

Data structures and types

Naming

Value processing logic

dailogs = [
    {
        "timestamp": "May 1, 2023 11:00:00",
        "text": "Hello, World!",
        "speaker": "Jack",
    },
    {
        "timestamp": "May 1, 2023 11:01:00",
        "text": "Hello, Copilot!",
        "speaker": "Copilot",
    },
]

# Parse the json object into `Dialog` object
...

Shortcuts

For Mac users, it is recommended to modify shortcuts related to the alt key, as the alt key with letters on Mac may be used by the input method. If you have customized the input method keylayout, ignore this sentence.

In addition, for commands without shortcuts, you can invoke the Command Palette and execute them by entering the query keyword after filtering.

Github Copilot

Other commands in Copilot

| Command | Description | Shortcut | Mac Shortcut | |-|:-|:-|:-| github.copilot.toggleCopilot| Enable/Disable Copilot completion suggestions | -| - github.copilot.collectDiagnostics| Collect diagnostics information | -| - github.copilot.openLogs| Open the log window | -| - github.copilot.sendFeedback| Open the community website | -| - github.copilot.signIn| Sign in | -| -

Github Copilot Chat

| Command | Description | Shortcut | Mac Shortcut | |-|:-|:-|:-| inlineChat.start| Code inline chat | - | - github.copilot.interactiveEditor.generate| Generate here (invoke the /generate function of inline chat at the cursor position) | - | - github.copilot.interactiveEditor.generateDocs| Generate documentation | - | - github.copilot.interactiveEditor.generateTests| Generate tests | - | - github.copilot.interactiveEditor.fix| Fix this | - | -

| Command | Description | Shortcut | Mac Shortcut | |-|:-|:-|:-| workbench.action.openChat.copilot| Open editor chat |-|-

Other commands in Copilot Chat

| Command | Description | Shortcut | Mac Shortcut | |-|:-|:-|:-| github.copilot.interactiveSession.feedback| Open github Issues |-|- github.copilot.debug.workbenchState| Log workbench state |-|- github.copilot.ghpr.applySuggestion| Provide code suggestions for Github Pull Request |-|-

Slash Commands tips in Copilot Chat

In the chat dialog, you can interact with Copilot Chat using commands starting with /.

Slash Commands Example

Slash Commands consist of four parts

Element	Description
- Agent	Specify the Agent, symbol is `@`, optional
- Commands	Specify the command, symbol is `/`, optional
- Variables	Reference content, symbol is `#`, optional
- User input command	optional

Example

/explain def helloworld():...

@vscode /api explain inlineChat.start

@workspace /explain def helloworld():...

Agent

Agent	Description
@vscode	vscode commands and plugin related questions
@workspace	Project workspace related questions

Inline Chat Slash Commands

Trigger inline chat using the command inlineChat.start and then use

Command	Description
/doc	Add documentation comments here
/explain	Explain the selected code
/fix	Fix the selected code
/tests	Generate unit tests for the selected code

Trigger using the command github.copilot.interactiveEditor.generate

Command	Description
/generate	Generate here, this command cannot be entered by the user

Alternatively, you can directly select the area and then enter the command in the inline chat to perform the command operation on the selected area.

Slide Chat Slash Commands

Trigger chat using the command workbench.action.chat.openInSidebar and then use
Or click the Copilot chat button on the sidebar
The Agent (environment) can also be specified in the Chat input box

Slash Commands

Command	Description
/api	Answer questions about vscode extension plugin development
/explain	Explain the selected code
/fix	Fix the selected code
/new	Create a new project workspace
/newNotebook	Create a new Jupyter Notebook
/terminal	Explain the commands in the command line
/tests	Generate unit tests for the selected code
/help	Help instructions
/clear	Clear the session

`/terminal` specific variables, starting with `#`

Only available in the /terminal command

Variable	Description
`#terminalLastCommand`	The last executed terminal command
`#terminalSelection`	The selected terminal command

Slash Commands in Quick Chat and Editor Chat

Quick Chat and Chat have the same Slash Commands
Editor Chat and Chat have the same Slash Commands

Parameter Settings

Open the VS Code command palette and enter Preferences: Open Settings to open the configuration file. Configure the relevant parameters in file mode.

You can find the complete parameter description in the package.json files of the copilot and copilot chat plugins in the extension marketplace.

Complete Configuration Reference

// settings.json
{   
    // ...
    "github.copilot.chat.welcomeMessage": "always",
    "github.copilot.chat.localeOverride": "auto",
    "github.copilot.editor.enableCodeActions": true,
    "github.copilot.editor.iterativeFixing": true,
    "github.copilot.editor.enableAutoCompletions": true,
    "github.copilot.enable": {
        "plaintext": false,
        "ini": false,
        "markdown": true,
        "*": true
    },
    "github.copilot.advanced": {
        "length": 4000,
        "inlineSuggestCount": 5,
        "top_p": 1,
        "temperature": "0.8",
        "listCount": 10,
        "stops": {
            "python": [
                "\ndef ",
                "\nclass ",
                "\nif ",
                "\n\n#"
            ],
            "*": [
                "\n\n\n"
            ]
        },
        "debug.showScores": true,
        "indentationMode": {
            "python": false,
            "javascript": false,
            "javascriptreact": false,
            "jsx": false,
            "typescript": false,
            "typescriptreact": false,
            "go": false,
            "ruby": false,
            "*": true
        }
    }
    // ...
}

Parameter Description

Proxy Parameters

Setting	Value Type	Description
`"http.proxy"`	string	Configure the network proxy address

Copilot Chat Parameters

Setting	Value Type	Description
`"github.copilot.chat.localeOverride"`	string	Set the Copilot language
`"github.copilot.chat.welcomeMessage"`	string	Whether to display a welcome message in Copilot Chat `first`: only on the first launch, `always`: always, `never`: never

Copilot Basic Parameters

Setting	Value Type	Description
`"editor.inlineSuggest.enabled"`	boolean	Enable inline suggestions
`"github.copilot.editor.iterativeFixing"`	boolean	Allow Copilot to provide iterative fixing suggestions
`"github.copilot.editor.enableAutoCompletions"`	boolean	Allow Copilot to provide auto completions
`"github.copilot.editor.enableCodeActions"`	boolean	Allow Copilot to provide code actions, which may include code refactoring, optimizing code structure, fixing errors, etc.

Setting File Types for Copilot

Setting	Value Type	Description
`"github.copilot.enable"`	json	Please put `"*": true` at the end. other language to disable Copilot, set `false` to disable, set `true` to enable

Copilot Advanced Parameters

github.copilot.advanced controls model parameters, which ultimately affect code generation. Its value is json.

|Setting |Value Type|Description | |:–|:–|:–| "length"| integer | Number of code words generated, default is 500 "top_p"| number | Control the range of model candidates, default value is 1, value range is 0.0~1.0 "temperature"| string | Control the creativity of the model, default value is "", the larger the value, the more unpredictable, value range is 0.0~1.0 "inlineSuggestCount"| integer | Number of inline suggestions, default is 3 "listCount"| integer | Control the number of suggestions in the Completions Panel, default is 10 "stops"| json | Control the stop signs when generating model code, can be controlled by language "indentationMode"| json | Specify whether the language adopts the indentation mode of that language, which may conflict with stops. For example, when using \{\} for indentation, comprehensive consideration is needed when setting this parameter. "debug.showScores"| boolean | Show the score of each suggestion in the code suggestion list

References

This site is open source. Improve this page.