The best VS Code extensions in 2026 are a lean, fast core rather than a marketplace haul. Every extension you install adds startup time and a little risk, so the goal is the smallest set that genuinely improves your day: a linter, a formatter, language support, version control help, and one AI assistant. Below is the set worth installing, a sample config to wire it together, and the categories you can safely skip.
The lean core worth installing
| Category |
What it does |
Why it earns a slot |
| Linter (e.g., ESLint) |
Flags bugs and bad patterns as you type |
Catches errors before you run anything |
| Formatter (e.g., Prettier) |
Auto-formats on save |
Ends style debates and diffs noise |
| Language support |
Syntax, IntelliSense per language |
Real autocomplete and type hints |
| Git enhancement (e.g., GitLens) |
Inline blame and history |
Understand who changed what and why |
| AI assistant |
Inline suggestions and chat |
Speeds boilerplate and explanations |
| EditorConfig |
Shared whitespace and EOL rules |
Keeps teams consistent across editors |
A sample format-on-save config
Wire the formatter to run automatically so you never think about it. Put this in your workspace settings file. Note the comment style uses double slashes, never a hash:
{
// format every file when you save it
"editor.formatOnSave": true,
// let the linter auto-fix on save too
"editor.codeActionsOnSave": {
"source.fixAll": "explicit"
},
// pick a default formatter so VS Code does not ask each time
"editor.defaultFormatter": "esbenp.prettier-vscode",
// keep startup fast by not auto-updating mid-session
"extensions.autoCheckUpdates": false
}
And a minimal lint setup script using shell comments:
// shell comments below use double slashes to stay valid as markdown
npm install --save-dev eslint prettier
npx eslint --init
How to choose extensions
- Start from empty and add on need. Install an extension when you hit a real friction, not preemptively.
- Prefer one tool per job. One linter, one formatter, one AI assistant. Duplicates fight each other.
- Check the maintenance signals. Recent updates, open issue responsiveness, and download counts tell you if it is alive.
- Disable per-workspace what you do not need everywhere. A Python extension does not need to load in a frontend project.
- Audit quarterly. Uninstall anything you have not consciously used in a month.
What to skip
- Theme and icon pack overload. One theme, one icon set. The rest is decoration that slows startup.
- Extensions duplicating built-ins. VS Code now has strong Git, search, and refactoring built in; do not re-add them.
- Abandoned extensions. No update in a year and a pile of open bugs means trouble. Move on.
- Three AI assistants at once. Pick one, learn its shortcuts, and stop. Stacking them just adds noise and cost.
If you are choosing between editors entirely, our VS Code versus Cursor comparison breaks down when an AI-first fork is worth switching to.
FAQ
How many VS Code extensions should I have?
As few as do real work, often under ten. Each one adds startup time and potential conflicts, so install on need rather than collecting them.
Do extensions slow down VS Code?
Yes, especially at startup and in large projects. The built-in extension performance tools let you see which ones cost the most load time so you can prune.
Are AI coding extensions worth it?
For most developers, one is genuinely useful for boilerplate and explanations. The value drops fast when you stack several; pick one and learn it well.
Should I sync my extensions across machines?
Yes. The built-in Settings Sync keeps your lean set and config consistent, which is far better than reinstalling by memory.
Where to go next
Tune the rest of your setup and skills: the best coding laptops to run it on, a VS Code versus Cursor breakdown, and how to debug code faster.