GitHub Copilot Setup and Best Practices for Daily Use
Get GitHub Copilot working in any editor, learn the prompts that actually produce good code, and avoid the habits that turn it into a liability.
GitHub Copilot was the first AI coding assistant most engineers used and it's still a solid daily driver. The newer chat and agent features have closed much of the gap with dedicated tools. This guide covers installation, configuration, and — more importantly — the habits that separate engineers who get value from Copilot from those who fight it.
Install
- VS Code: install the GitHub Copilot extension from the Marketplace, sign in with your GitHub account.
- JetBrains: install from the plugin marketplace and authenticate via the IDE.
- Neovim: use
copilot.vimor the LSP-basedcopilot.luafor Lazy users.
Subscription
There's a free tier with limits, an Individual plan, and a Business plan with stronger privacy controls (your code is never used to train models). Pick the tier that matches your employer's policies — most companies require Business.
Three modes
- Inline completions: ghost-text suggestions as you type, accepted with Tab.
- Copilot Chat: a chat panel inside the IDE for explanations, fixes, and small refactors.
- Copilot Agents (Workspace): longer-running tasks that span multiple files, similar to Cursor's Composer.
Patterns that produce good code
- Write the function name and a docstring first. Copilot generates much better implementations when it has the contract.
- Open the related files. Suggestions improve when other relevant files are in your tab list.
- Be willing to reject. Tabbing through bad suggestions is a habit that leads to silent bugs.
Habits that hurt
- Accepting suggestions you didn't read line by line.
- Letting Copilot guess at security-sensitive code (auth, crypto, SQL builders).
- Skipping tests because "the AI wrote it."
Configuration tips
- Add a
.copilotignoreso secrets and lockfiles never appear in the context window. - Disable Copilot in markdown if it autocompletes paragraphs you didn't ask for.
- Set up custom prompts in Copilot Chat for repeatable tasks like "explain this PR" or "write tests for this file."
Copilot vs the alternatives
Copilot is the most polished IDE-native experience and has the best multi-editor support. Claude Code wins on autonomous tasks, Cursor wins on multi-file edits inside the editor, and Codeium wins on price for individual developers. If you already pay for GitHub, Copilot is the lowest-friction option.
Continue reading
Related essays
Continue.dev: The Open-Source Copilot Alternative You Can Customize
Continue is the open-source AI coding extension for VS Code and JetBrains. It works with any model — including local ones — and you control every part of the pipeline.
Codeium Setup and an Honest Review After 30 Days
Codeium is the free-tier-friendly Copilot alternative. Here's how to install it, what it does well, and where it falls short on real-world coding work.
Claude vs ChatGPT vs Gemini: Which AI for Coding in 2026?
An honest, hands-on comparison of the three big assistants for daily software engineering work — context windows, code quality, agent capabilities, and where each one wins.