What VS Code is
Visual Studio Code is a free, open-source code editor from Microsoft that has become the default text-editing environment across most of computing. It edits plain text files, and everything beyond that — language support, LaTeX compilation, notebook execution, Git integration — arrives through extensions you install.
The reason it matters to researchers rather than only to developers is consolidation. A typical research setup involves LaTeX in one application, Python in another, Markdown notes in a third, and Git in a fourth, each with its own shortcuts and its own quirks. VS Code collapses that into one window with one set of keybindings, because all four are ultimately plain-text files. Learning it once and using it for everything is the single biggest reduction in tool friction available to a computational researcher.
Why researchers use it
- LaTeX with SyncTeX — the LaTeX Workshop extension gives auto-compile on save, live PDF preview, and click-in-the-PDF-to-jump-to-source, which is the feature you miss most in every other editor.
- Notebooks natively —
.ipynbfiles open and run in the editor, with variable inspection, so you are not switching to a browser for exploratory work. - Remote development — edit files on an HPC cluster or a lab server over SSH as if they were local. For anyone running jobs on shared infrastructure this is transformative.
- Git built in — stage, diff, commit, and resolve conflicts without leaving the file you are editing.
- It is the same editor everywhere — your configuration syncs across machines, and the same shortcuts work on your laptop, the lab workstation, and a remote node.
Where it fits in a research workflow
VS Code sits at the centre of a plain-text research workflow, which is the point. Your LaTeX manuscript, your Python analysis, your Markdown notes, and your .bib file are all text, all in one Git repository, all editable in one place — the arrangement described in Git and GitHub for researchers and the reason to prefer text formats in the first place.
Concretely: write in LaTeX against a .bib file exported from Zotero or maintained in JabRef; run analysis in a uv or conda environment; render documents with Quarto or Pandoc; commit to GitHub. The full LaTeX Workshop setup covers the one genuinely fiddly part. Where it hands off: RStudio is still better for R, and PyCharm is better for a large, structured Python codebase.
Getting started
An hour, and install extensions one at a time.
- Install VS Code, then add extensions for what you actually do today — Python and Jupyter, or LaTeX Workshop, or Markdown. Installing twenty at once produces a confusing editor and no understanding of which one is misbehaving.
- For LaTeX, install a local TeX distribution first (TeX Live or MiKTeX), then LaTeX Workshop, then set your build recipe to match your bibliography tool. This is the step that takes the hour.
- Open a folder, not a file. VS Code is organised around project folders, and half its features do nothing until it can see one.
- The step people skip: learn the command palette. One keystroke, type what you want, and every command in the editor is searchable — it replaces memorising menus and is the difference between using VS Code and fighting it.
VS Code vs the alternatives
| Alternative | Does it better | Pick it if |
|---|---|---|
| RStudio | R specifically — plots pane, environment inspector, knitting | R is your primary language |
| PyCharm | Deep refactoring and debugging on large Python codebases | The codebase is the product, not the notebook |
| TeXstudio | LaTeX out of the box with no configuration | You only write LaTeX and resent setup |
| Overleaf | Real-time collaborative LaTeX in a browser | Co-authors will not install anything |
VS Code wins on breadth and loses on depth in each individual domain. If you do three of these things, it is the right answer; if you do one, the specialist may be.
Cost, licensing, and your data
Free. The Microsoft-branded builds are released under a proprietary licence over the MIT-licensed source; fully open builds (VSCodium) exist if your institution requires them. Most extensions are free and open source, though some AI assistants and language servers are commercial.
Two things to be deliberate about. Microsoft-branded builds send telemetry by default — it is documented and it can be disabled, and some institutions require that it be. And extensions run with your permissions and can read your files: the marketplace has had malicious and typosquatted extensions, so check the publisher and install count before adding something obscure. Neither is a reason to avoid VS Code; both are reasons to configure it deliberately on a machine that touches sensitive data.
The honest review
Strengths. The consolidation is real and compounding. One editor, one set of shortcuts, one configuration, applied to your manuscript, your analysis, your notes, and your remote cluster session. Add remote SSH development and it becomes the tool that makes working on shared infrastructure feel normal instead of painful — which for anyone with an HPC account is worth more than every other feature combined.
Limitations. It is an editor that becomes an IDE only after you configure it, and the LaTeX setup in particular is an hour of build recipes and path variables that specialist editors simply do not require. Extension quality varies and the marketplace has real supply-chain risk. It uses more memory than its reputation suggests once several extensions are loaded. And its notebook support, while good, still lags a dedicated JupyterLab session for heavy interactive work.
Verdict. Adopt it if you work across more than one plain-text format — the consolidation pays for the setup within a month. Skip it if you write only R (use RStudio) or only LaTeX and hate configuration (use TeXstudio). The condition that flips the answer is whether you have an SSH account on a cluster: if you do, remote development alone justifies the switch.
Common questions
Is VS Code good for LaTeX?
Yes, with the LaTeX Workshop extension: auto-compile, live preview, and two-way SyncTeX navigation. It takes about an hour to configure the first time, which is the honest trade against TeXstudio or Overleaf working immediately. The full setup guide covers it.
Can I run Jupyter notebooks in VS Code?
Yes — .ipynb files open natively with the Jupyter extension, run against any kernel you have installed, and include variable inspection. For heavy interactive work a full JupyterLab session is still smoother, but for most analysis this is enough and keeps everything in one window.
Is VS Code better than PyCharm?
For mixed work — Python plus LaTeX plus Markdown plus remote sessions — yes, because it does all of it in one place. For a large, structured Python codebase where refactoring and debugging depth matter, PyCharm is the stronger tool. Many researchers use VS Code daily and open PyCharm for the occasional large refactor.