CopilotTyping all CRUD routes by hand
A full set of CRUD routes that would take 2–3 hours to type is done in 20 minutes with consistent style and no copy-paste errors.
Track 13 · Advanced
13.2.5
GitHub Copilot
AI pair programmer built into your editor — writes code as you type
1Overview
GitHub and OpenAI's inline completion tool — its core strength is fast, in-the-moment suggestions tuned to data-science workflows like pandas, numpy and ggplot2, with a chat panel and Agent Mode layered on top for bigger tasks.
GitHub Copilot, from GitHub (Microsoft) and OpenAI, sits directly inside your code editor. As you type it suggests the next line, function, or whole block — press Tab to accept. A chat panel explains errors and suggests refactors, and Agent Mode can plan, write, and run multi-step tasks across your project files. → Verified students get unlimited completions free via the GitHub Student Developer Pack.
GitHub Copilot is an AI pair programmer built into your editor — it writes code as you type. Below: what it's best at, and what to watch for.
1.1Before you start
AI coding assistants— Reading AI coding assistants shows how in-editor tools collaborate on your existing codebase before exploring GitHub Copilot.A repo you can push to— First, work in GitHub—store repositories, handle pull requests, Issues, and Actions, where AI agents already convert Issues into completed PRs.
1.2After this chapter you can
→Accept inline suggestions as you type, in VS Code, JetBrains, or Xcode
→Use Copilot Chat to explain an error or refactor a function
→Delegate a multi-file task to Agent Mode
1.3Best for
Fast code-as-you-type completion in Python or R, especially with pandas, numpy, ggplot2, and Bioconductor workflows already understood.
1.4Watch out
Needs an editor install first (VS Code, JetBrains…) — no browser interface — and can hallucinate plausible-but-wrong API calls; always verify.
1.5Free vs paid
Free tier: 2,000 completions + 50 chat requests/month. Pro $10/user/mo unlimited; Business $19/seat/mo; Enterprise $39/seat/mo. Verified students get it free.
2Lessons 5
2.1 Install the GitHub Copilot extension in VS Code
The official GitHub Copilot extension that adds AI‑powered code suggestions to Visual Studio Code.
You will have the extension installed and ready to activate inside your editor.
- Open VS Code.
- Open the Extensions view (Ctrl+Shift+X).
- Search for "GitHub Copilot".
- Click Install on the GitHub Copilot extension entry.
- Reload VS Code when prompted.
- You'll see The extension appears in the installed list and a Copilot icon shows in the status bar.
- Takeaway Installing an IDE extension is the first step to bring external AI services into your local development workflow.
2.2 Enable GitHub Copilot for your account
The activation process that links the installed extension with a Copilot plan (Free, Pro, or Student).
You will be signed in and have Copilot enabled so it can start providing suggestions.
- Click the Copilot icon in VS Code’s status bar.
- Select "Sign in to GitHub" and complete the OAuth flow.
- After signing in, choose a plan (Free or Student if verified).
- Confirm activation when prompted.
- Verify that the status bar now shows “Copilot: Enabled”.
- You'll see A green check‑mark next to the Copilot icon indicating it is active for the current workspace.
- Takeaway Linking the extension to a licensed plan unlocks the AI model behind the suggestions.
2.3 Generate code suggestions with inline completion
Copilot’s core feature that suggests the next line, function, or block as you type.
You will receive and accept an AI‑generated suggestion inside a source file.
- Open a new JavaScript file in VS Code.
- Start typing a function signature, e.g.,
function greet(name) {. - Pause; Copilot will display a grayed suggestion for the next line.
- Press Tab to accept the suggestion.
- Continue typing to trigger further suggestions and accept as desired.
- You'll see The editor inserts completed code that matches the context you typed, and the cursor moves to the next editable position.
- Takeaway Inline completions accelerate routine coding by predicting syntactically correct snippets based on surrounding code.
2.4 Use Copilot Chat to ask for explanations
An interactive chat panel where you can type natural‑language queries and receive AI‑generated responses.
You will open the chat, ask a question about your code, and view the answer.
- In VS Code, open the Copilot Chat panel via the command palette (Ctrl+Shift+P) → “Copilot: Open Chat”.
- Type a question such as "What does this function do?" referencing the code you just wrote.
- Press Enter to submit the query.
- Read the response that explains the function’s behavior.
- Optionally click any suggested edit link to apply changes.
- You'll see A chat window displays a textual explanation of your code and, if applicable, inline edit suggestions.
- Takeaway Chat mode turns the AI into an on‑demand documentation assistant, helping you understand or improve existing code.
2.5 Run a multi‑step task with Copilot Agent Mode
Agent Mode lets Copilot plan, write, and execute a sequence of actions across project files.
You will initiate an agent to create a new utility file and add it to the repository.
- In the Copilot Chat panel, type: "Create a new file utils.js that exports a function sum(a,b) returning their sum, then add an import for it in index.js".
- Press Enter to send the request.
- Review the agent’s step‑by‑step plan displayed in chat.
- Confirm each step by clicking the provided “Execute” button (or type "yes").
- After execution, open utils.js and index.js to verify the new file and import were added.
- You'll see Two files are created/modified: utils.js contains the sum function, and index.js includes an import statement referencing it.
- Takeaway Agent Mode extends Copilot from single‑line suggestions to orchestrating larger code changes across a project.
3You’ll know it worked 14 checkable outcomes in this chapter
- ✓All CRUD routes for a resource are present and consistent without manual copy-paste
- ✓Ingestion pipeline processes dates correctly without errors after applying new regex
- ✓Two new modules (loader.py, plotter.py) and a test_loader.py import the loader module
- ✓The command lists each .fasta file and its sequence count without errors
- ✓Function returns a dictionary of amino acid counts when called with a valid file path
- ✓The refactored function passes all tests and Copilot lists each change
- ✓The pull request shows commits authored by the assigned AI agent ready for review
- ✓The CLI outputs a summary of agents and target files with no errors before any changes are made
14 outcomes in all — one per recipe below.
4FAQ, Tips & How-to 39
one problem, one solution, one action
Internal tools & ops7
One big analysis script needs modules, type hints, and a loader test
One natural-language request spans changes that would otherwise mean editing several files by hand in sequence.
Code throws a KeyError
Selecting the code first scopes the chat's context to it, instead of guessing from the whole file.
Need to count sequences in each .fasta file
Reading the generated command before running it catches a wrong flag before it touches your files.
Want an issue turned into code you can review
The work happens on GitHub's own infrastructure, so you review a finished PR rather than watch it edit.
Empty function with a detailed comment
Specificity in the comment — inputs, format, sort order — narrows what Copilot guesses at, not just the goal.
Need a function rewritten and explained
Requesting an explanation alongside the refactor turns the diff into something you can verify, not just trust.
Research & data tools2
A regex bug that stalled the ingestion pipeline for two days is identified and fixed in 30 minutes with a clear explanation the whole team can follow.
Need nonstop EMG readings at 2 kHz
Both EMG channels stream continuously into the ring buffer with no missed samples at 2 kHz; the buffer hand-off to the signal-processing task is interrupt-safe without a RTOS mutex.
When you type, Copilot shows faint gray text (ghost text) that can be accepted or ignored
Need to turn AI suggestions into actual code quickly
Pressing Tab inserts the suggested ghost text as real code, letting you build programs line-by-line
Write a detailed comment before a function
Starting the function definition after a detailed comment lets Copilot suggest the body, next lines, or whole blocks
Want a quick way to accept suggested code
Pressing Tab accepts the suggested line or block, allowing you to build the function piece by piece
You can launch the in-editor Chat without leaving your IDE
You can keep the dialogue open to ask for clarifications, simpler code, or added comments
Need to work on several source files in one go
Lets Copilot plan, write, run, and iterate over several project files in one request
Want to approve or reject every AI edit
Gives you final authority to accept or reject every modification Copilot made
A quick test validates that Copilot's numerical output matches known results, catching errors early
You can open the Copilot CLI directly in your terminal and it will confirm you are logged in before accepting tasks
Typing `?` shows the built-in help menu so you can discover available shortcuts and usage tips without leaving the terminal
Can’t tell AI which file to use
Using `@` before a filename tells Copilot which existing file the command should act on
Want to execute a suggested command instantly
Prefixing a suggestion with `/` lets you execute the generated command immediately from the CLI
Always review each generated command before executing it to avoid unintended side effects
Need to locate all FASTA files and see how many sequences each contains
You can ask Copilot to produce a concrete shell pipeline for a specific file-processing task
GitHub groups AI agents with human teammates, making them selectable for any issue
Review AI‑generated code changes
After assignment, the coding agent proposes code changes that you review and merge just like a human teammate
What is GitHub Copilot and what is it for?
GitHub Copilot is an AI pair programmer that lives inside your code editor. As you type, it suggests the next line or whole blocks of code as faint grey ghost text, which you accept with the Tab key. Beyond autocomplete, it has a chat panel that explains errors and refactors code in plain language, and an Agent Mode that can plan and edit across multiple files. It is built for writing code faster, especially everyday Python and R data work, turning a plain-English comment into working code without leaving your editor. It works in VS Code, JetBrains IDEs, Visual Studio, and Xcode.
Does Copilot send my code to the cloud? What about privacy?
Yes, Copilot is a cloud service. Your prompts and surrounding code context are sent to GitHub and the underlying model providers to generate suggestions, so it is not a fully local tool. For most coursework and open-source code this is fine, but you should not paste confidential patient data or unpublished research into it without checking your institution's policy first. GitHub offers controls such as a content exclusion setting to keep specified files out of Copilot, and Business and Enterprise plans add data handling and policy controls. Review GitHub's privacy and trust documentation to understand exactly what is collected and retained before using it on sensitive material.
Can I use Copilot in the terminal, not just the editor?
Yes, through the GitHub Copilot CLI. You start it in your terminal, and after confirming you are logged in it waits for a task. Describe a command or job in plain English, such as finding every .fasta file under a folder and counting the sequences in each, and Copilot can write, test, and debug code right from the terminal. You can mention files with the @ symbol and run / commands, and you should read each suggested command before running it. The CLI is handy for glue work like file wrangling, quick scripts, and git steps without switching to the editor. Like chat and agents, CLI features consume AI Credits.
How does GitHub Copilot compare to Cursor?
Both put AI into a familiar VS Code-style editor, but they sit at different points. Copilot is an extension you add to an editor you already use (VS Code, JetBrains, Visual Studio, Xcode), and its strongest feature is fast inline autocomplete, with chat and Agent Mode layered on top. Cursor is a standalone AI-first editor (a VS Code fork) built around agentic, multi-file changes and deep project context as the default way of working. Copilot has a more generous always-free tier (2,000 completions a month) and free access for verified students through GitHub Education. Choose Copilot to add AI to your current editor; choose Cursor if you want an editor built around the agent.
How do I install GitHub Copilot and get started?
The most common path is through VS Code. Install VS Code, open the Extensions panel, search for the GitHub Copilot extension, install it, and sign in with a GitHub account. Copilot is also available in JetBrains IDEs, Visual Studio, and Xcode. The Free plan needs no credit card. Once installed, open a .py file and type a comment describing what you want, for example loading a CSV and plotting a heatmap. Copilot reads the comment and proposes the next lines as ghost text. Press Tab to accept a suggestion or keep typing to ignore it. There is no browser or standalone app, so the one-time setup is installing an editor and the extension.
Can students get GitHub Copilot for free?
Yes. Verified students get free access to Copilot and its premium features through GitHub Education. To qualify you apply for GitHub Education and verify your student status, typically with a school-issued email address and proof of enrollment. Verified teachers and maintainers of popular open-source projects also get Copilot Pro for free. This makes Copilot one of the cheapest ways for a student to get a full-featured AI coding assistant, since you can use it across VS Code and other supported editors without paying.
How do I actually write code with Copilot? What is the most common task?
The everyday workflow is comment-driven autocomplete. Open a code file and write a specific comment naming the input, the operation, and the desired output, for example a comment that says load gene_counts.csv, normalize by library size, and plot a heatmap of the top 50 variable genes. Then start a function signature on the next line. Copilot completes the body as grey ghost text. Press Tab to accept good lines and type over the ones you do not want. The more concrete your comment, the closer the suggestion lands. Refine the comment and re-trigger if the suggestion drifts. This same Tab-to-accept flow works across VS Code, JetBrains, Visual Studio, and Xcode.
What is Copilot Chat and when should I use it?
Copilot Chat is a conversation panel inside your editor for the understanding that inline autocomplete skips. Open it from the Copilot icon, select a block of code so the chat has context, and ask in plain language. It is ideal for explaining an error such as a KeyError, suggesting a refactor, answering why does this code work, or rewriting a loop into a cleaner pandas operation. The answer comes back as an explanation plus code you can apply, and you can iterate with follow-up questions in the same thread. Unlike completions, chat does consume AI Credits, so on the Free plan chat usage is one of the first things to run out.
What is Agent Mode and how is it different from autocomplete?
Agent Mode handles tasks too big for line-by-line autocomplete. In the Copilot Chat panel, switch the mode dropdown to Agent, then send a multi-step task such as splitting one script into a data-loading module and a plotting module with type hints and a test. Copilot plans the work, edits across several files, and can run code to check itself. It shows a summary of the files it changed with Keep and Undo controls so you review every edit before accepting. You stay the reviewer. Agent Mode consumes AI Credits and a deep multi-file run costs more than a single chat, so it draws on your plan's monthly Credit allowance.
Can I choose which AI model Copilot uses?
Yes. The Chat panel has a model picker, and even the Free plan gives access to a selection of models, including ones from Anthropic (Claude) and OpenAI (GPT). Premium and more powerful models are available on the paid tiers such as Pro+ and the higher individual plans, and using a premium model in chat draws on your AI Credits. A practical approach is to use a stronger model for harder refactors and ask Copilot to explain each change so you can confirm it. Model availability changes over time, so check the picker in your editor for the current list.
Agents keep overwriting each other’s files
Create a markdown file (plan.md) that maps each target component or directory to a specific sub‑agent. By being explicit about which files each agent may modify, you prevent overlapping writes and merge conflicts during parallel execution.
Need to check a new fleet launch safely
Use the `--dry-run` (or similar) flag when first launching the fleet. The command parses plan.md, checks that all referenced paths exist, and simulates agent startup without writing any code, catching typos or mis‑assignments early.
Many separate code changes need checking
After the fleet finishes, it creates separate Git branches each containing only that agent’s final diff. By reviewing each branch individually you retain full control over integration and can catch logical errors before merging to main.
The same set on /recipes, filtered by tool and role.
5Videos 3
Master GitHub Copilot in One Video | Complete 2026 Walkthrough 🚀
The all-in-one beginner overview if you'd rather watch one video than a series.
How to use agents, skills, and instructions in Copilot CLI | Tutorial for beginners
Official and current. Watch to see Copilot as a terminal agent, not just inline autocomplete.
Hands-On Lab: The Power of GitHub Copilot in VS Code
Straight from the VS Code team. The deepest official walkthrough — do it as a lab with VS Code open.
6FAQ 10
What is GitHub Copilot and what is it for?
GitHub Copilot is an AI pair programmer that lives inside your code editor. As you type, it suggests the next line or whole blocks of code as faint grey ghost text, which you accept with the Tab key. Beyond autocomplete, it has a chat panel that explains errors and refactors code in plain language, and an Agent Mode that can plan and edit across multiple files. It is built for writing code faster, especially everyday Python and R data work, turning a plain-English comment into working code without leaving your editor. It works in VS Code, JetBrains IDEs, Visual Studio, and Xcode.
Does Copilot send my code to the cloud? What about privacy?
Yes, Copilot is a cloud service. Your prompts and surrounding code context are sent to GitHub and the underlying model providers to generate suggestions, so it is not a fully local tool. For most coursework and open-source code this is fine, but you should not paste confidential patient data or unpublished research into it without checking your institution's policy first. GitHub offers controls such as a content exclusion setting to keep specified files out of Copilot, and Business and Enterprise plans add data handling and policy controls. Review GitHub's privacy and trust documentation to understand exactly what is collected and retained before using it on sensitive material.
Can I use Copilot in the terminal, not just the editor?
Yes, through the GitHub Copilot CLI. You start it in your terminal, and after confirming you are logged in it waits for a task. Describe a command or job in plain English, such as finding every .fasta file under a folder and counting the sequences in each, and Copilot can write, test, and debug code right from the terminal. You can mention files with the @ symbol and run / commands, and you should read each suggested command before running it. The CLI is handy for glue work like file wrangling, quick scripts, and git steps without switching to the editor. Like chat and agents, CLI features consume AI Credits.
How does GitHub Copilot compare to Cursor?
Both put AI into a familiar VS Code-style editor, but they sit at different points. Copilot is an extension you add to an editor you already use (VS Code, JetBrains, Visual Studio, Xcode), and its strongest feature is fast inline autocomplete, with chat and Agent Mode layered on top. Cursor is a standalone AI-first editor (a VS Code fork) built around agentic, multi-file changes and deep project context as the default way of working. Copilot has a more generous always-free tier (2,000 completions a month) and free access for verified students through GitHub Education. Choose Copilot to add AI to your current editor; choose Cursor if you want an editor built around the agent.
How do I install GitHub Copilot and get started?
The most common path is through VS Code. Install VS Code, open the Extensions panel, search for the GitHub Copilot extension, install it, and sign in with a GitHub account. Copilot is also available in JetBrains IDEs, Visual Studio, and Xcode. The Free plan needs no credit card. Once installed, open a .py file and type a comment describing what you want, for example loading a CSV and plotting a heatmap. Copilot reads the comment and proposes the next lines as ghost text. Press Tab to accept a suggestion or keep typing to ignore it. There is no browser or standalone app, so the one-time setup is installing an editor and the extension.
Can students get GitHub Copilot for free?
Yes. Verified students get free access to Copilot and its premium features through GitHub Education. To qualify you apply for GitHub Education and verify your student status, typically with a school-issued email address and proof of enrollment. Verified teachers and maintainers of popular open-source projects also get Copilot Pro for free. This makes Copilot one of the cheapest ways for a student to get a full-featured AI coding assistant, since you can use it across VS Code and other supported editors without paying.
How do I actually write code with Copilot? What is the most common task?
The everyday workflow is comment-driven autocomplete. Open a code file and write a specific comment naming the input, the operation, and the desired output, for example a comment that says load gene_counts.csv, normalize by library size, and plot a heatmap of the top 50 variable genes. Then start a function signature on the next line. Copilot completes the body as grey ghost text. Press Tab to accept good lines and type over the ones you do not want. The more concrete your comment, the closer the suggestion lands. Refine the comment and re-trigger if the suggestion drifts. This same Tab-to-accept flow works across VS Code, JetBrains, Visual Studio, and Xcode.
What is Copilot Chat and when should I use it?
Copilot Chat is a conversation panel inside your editor for the understanding that inline autocomplete skips. Open it from the Copilot icon, select a block of code so the chat has context, and ask in plain language. It is ideal for explaining an error such as a KeyError, suggesting a refactor, answering why does this code work, or rewriting a loop into a cleaner pandas operation. The answer comes back as an explanation plus code you can apply, and you can iterate with follow-up questions in the same thread. Unlike completions, chat does consume AI Credits, so on the Free plan chat usage is one of the first things to run out.
What is Agent Mode and how is it different from autocomplete?
Agent Mode handles tasks too big for line-by-line autocomplete. In the Copilot Chat panel, switch the mode dropdown to Agent, then send a multi-step task such as splitting one script into a data-loading module and a plotting module with type hints and a test. Copilot plans the work, edits across several files, and can run code to check itself. It shows a summary of the files it changed with Keep and Undo controls so you review every edit before accepting. You stay the reviewer. Agent Mode consumes AI Credits and a deep multi-file run costs more than a single chat, so it draws on your plan's monthly Credit allowance.
Can I choose which AI model Copilot uses?
Yes. The Chat panel has a model picker, and even the Free plan gives access to a selection of models, including ones from Anthropic (Claude) and OpenAI (GPT). Premium and more powerful models are available on the paid tiers such as Pro+ and the higher individual plans, and using a premium model in chat draws on your AI Credits. A practical approach is to use a stronger model for harder refactors and ask Copilot to explain each change so you can confirm it. Model availability changes over time, so check the picker in your editor for the current list.
7Glossary 13 terms
Show the 13 terms
GitHub Copilot
.py- The file extension for a Python source code file — any file ending in .py contains Python instructions the computer can run.
def- A Python keyword that marks the start of a new function definition — everything indented beneath it is the code that runs when you call that function.
Counter- A Python built-in class (from the `collections` module) that counts how many times each item appears in a sequence and stores the results as a dictionary.
DataFrame- A table of data provided by the pandas library — it has named columns and numbered rows, similar to a spreadsheet you can manipulate with code.
KeyError- A Python error that occurs when you try to access a dictionary or DataFrame column using a name that does not exist in it.
ValueError- A Python error that occurs when a function receives a value of the right type but an inappropriate content — for example, passing an empty list where data is required.
groupby- A pandas operation that splits a table into groups based on the values in a column (like splitting experiment rows by treatment condition) so you can calculate statistics for each group separately.
.div()- A pandas method that divides every value in a Series or DataFrame by a number or another Series — commonly used to convert raw counts into proportions.
pd.read_csv- A pandas function that reads a CSV (comma-separated values) file from disk and loads it into a DataFrame so you can analyse it with Python.
sns.heatmap(...)- A seaborn function that draws a colour-coded grid where each cell's colour represents a numeric value — useful for spotting patterns across many samples or genes at once.
method- A function that belongs to an object or class — for example, `.mean()` is a method on a pandas DataFrame that calculates the average of a column.
@- In GitHub Copilot Chat (VS Code), typing @ opens a list of chat participants — domain-expert agents such as @workspace or @github — that you can direct your question to for specialised help.
/- In GitHub Copilot Chat, typing / opens a list of slash commands — shortcuts for common tasks like /explain (explain selected code) or /tests (generate unit tests) — so you do not need to write a full prompt.
8See also
💬 Discuss this chapter
Ask, share, or report — over on the Heidelberg AI community forum.