Suzuri
Reading

PDF Viewer

A native PDF pane with continuous scroll, zoom, and real text selection, that reloads in place when the file changes on disk.

Open a .pdf and it opens in a Suzuri tab, not in Preview or a browser. That matters because the PDF is the other half of the work: the paper you're citing, the draft you just compiled, the figure you're checking.

  • Continuous scroll through the whole document, not a page at a time
  • Real text selection — glyph-accurate, so copied text is the document's text, not OCR
  • Zoom, fit-to-width, and actual size
  • Reloads in place when the bytes change on disk, keeping your scroll position — which is what makes typeset preview feel live

It works on remote projects too: a PDF rewritten on an SSH host reloads in the local pane.

Keybindings

ActionmacOS
Zoom inCmd+=
Zoom outCmd+-
Reset zoomCmd+0
Actual size (100%)Cmd+1
Fit to view widthCmd+Shift+0
Copy all document textCmd+C

On Windows and Linux, substitute Ctrl. The same actions are available from the command palette as pdf_viewer: zoom in, pdf_viewer: fit to view, and so on, so you can rebind them.

Copy all document text

Cmd+C with no selection copies the entire document's text — handy for pasting a paper into an agent thread.

Getting to a PDF

  • Click a .pdf in the project panel
  • Follow a [](...) link to an attached PDF
  • Put the cursor on a cite key and press Cmd+Alt+O — see Citations
  • Preview a .typ or .tex file — see Typesetting

Credits

The viewer is built on hayro, a pure-Rust PDF renderer by Laurenz Stampfl. It was originally written by David Turnbull as a contribution to Zed (zed#51040), closed there as out of scope for a code editor, and adopted into Suzuri — where a research editor makes it a core surface. The glyph-accurate text extraction comes from the same author's hayro work.

On this page