Learn · Obsidian for researchers
Obsidian starts as a blank slate. A handful of free community plugins turn it into a research environment — a place to hold your literature, structure your data, and write from it. Here are the ones that earn their place: what each does, why it matters for research, and how to switch it on. All free, all local.
The shortlist:
| Plugin | What it does | Why for research |
|---|---|---|
| Zotero Integration | pulls citations, PDFs, and highlights from Zotero into notes | turns reading into atomic literature notes |
| Dataview | queries your notes' metadata like a database | reading lists, codebooks, status dashboards |
| Templater | dynamic note templates | consistent structure every plugin can read |
| Longform | compiles many notes into one manuscript | write a thesis in atomic pieces |
| Canvas (built-in) | a spatial board of live notes | evidence maps and argument structure |
Once, per plugin: Settings → Community plugins → Browse, search the name, Install, then Enable. That is the whole process. Built-in features like Canvas and the graph view need no install — they are already there.
Before any plugin, three things Obsidian already does are the backbone of a research vault:
[[brackets]] and every note that mentions it becomes connected. This is the whole idea; the plugins build on it.If you already keep references in Zotero, this plugin is the bridge from "I read it" to "I thought about it". It imports a paper's metadata, your PDF annotations, and your highlights into an Obsidian note — where you then rewrite the idea in your own words, one atomic note per concept, each one linked back to its source citation so the idea never loses track of where it came from.
Why it matters: it keeps the two tools in their lanes. Zotero stores the source — the PDF, the metadata, the bibliography. Obsidian holds your thinking. The plugin moves your highlights across so you are never retyping, and the atomic notes you write from them are what actually connect across papers in the graph. Keeping each atomic note linked to its citation is the quiet payoff: months later, when that note becomes a sentence in your manuscript, it already carries its source — you are never hunting for where you read something. Pair it with Better BibTeX (a Zotero add-on) for stable citation keys that never break.
How to use it: install and enable it, point it at your Zotero library, and set an import template (the plugin ships sensible defaults). New to Zotero itself? Start with our Zotero for researchers guide.
Dataview reads the metadata at the top of your notes (the YAML frontmatter) and turns it into live tables and lists: a reading list filtered by status, a codebook dashboard, every paper from a given year. You write a short query once and it updates itself as your notes change.
The JSON-drift trap. A common mistake is to keep structured research data — a table of participants, a coding scheme, a reading log — in a hand-edited JSON file inside the vault. It drifts: keys get inconsistent, a typo breaks a field, and the file quietly falls out of sync with your notes. The fix is to stop maintaining a separate data file at all. Keep the data in each note's frontmatter, where it lives next to the thing it describes, and let a query layer read it. That gives you one source of truth that cannot drift.
Which query layer depends on how your head works. Dataview uses a friendly query language over frontmatter. Bases (now built into Obsidian) gives a no-code, spreadsheet-like view of the same fields. Datacore is a faster, interactive successor to Dataview. And if you think in SQL, SQLSeal runs real SQL queries over your CSV and JSON files. Any of them beats a hand-kept JSON file — the point is to query a single source, not to edit data by hand in two places.
Templater inserts dynamic templates: today's date, prompted fields, a fixed heading structure. Every new interview note, literature note, or codebook entry starts identical. Why it matters: Dataview and a codebook only work if your notes share a shape. Templater is what guarantees that shape without you retyping it.
Longform assembles many small notes into a single manuscript you can reorder by dragging. Why it matters: you draft in atomic, linkable pieces the way Obsidian is good at, then compile to one document. When it is time for formal formatting, export the compiled text through Pandoc to Word or LaTeX.
Do not install thirty plugins. Each one is code that can break on an update, and a vault stuffed with half-used plugins is slower and more fragile than a lean one. Start with Zotero Integration, Dataview, and Templater, add Canvas and Longform when a real need appears, and stop there until something specific pushes you further. Excalidraw (hand-drawn models) and Tasks (tracking #needs-citation follow-ups) are worth knowing about for when that need arrives.
You can assemble this set yourself — install the three, wire up your templates, and build your dashboards.
If you work with qualitative data and would rather start from a vault where Dataview, Templater, and Canvas are already installed and wired into a codebook, evidence map, and audit trail — QualiVahti Local is that set, configured, with a worked demo study inside.
See the pre-configured vault — €49The set: Obsidian for researchers — 5 questions no guide answers, build a qualitative codebook, and Canvas as an evidence map. More on the Learn page.