Suppr Zotero Plugin: a context menu in front of a hosted translator

A Zotero plugin that ships a window into a paid service

Reading a paper in a language you do not know means either finding a translator and reattaching the result yourself, or doing nothing at all. This plugin removes the first half of that work. You pick a PDF in your library, open the right-click menu, and the vendor's translation service takes the file from there.

Who gains from that is narrow and clear: researchers with large PDF libraries who read outside English and who are willing to let a hosted service see the file. What comes back is a translated attachment on the same Zotero entry. There is no offline mode, no endpoint you can point at your own server, and no quota that lives outside the Suppr account, where new users are credited with a quarter of a million Chinese characters or one million English letters.

Right-click translation uploads the PDF and attaches a second file

The whole operation is four steps: select the PDF in the Zotero library, right-click it, choose the Suppr translation entry, and wait. Three things happen during that wait. The document is uploaded automatically, processed on the vendor's side, and the finished file is downloaded and attached to the original item.

Two consequences follow from a flow that returns a new attachment rather than editing the old one. Your original PDF stays exactly where it was, untouched. And you then hold two files on one entry, with no in-place reading view, no side-by-side pane, and no translated text pushed into the abstract or notes fields. One more wrinkle: the menu label in the documentation is written in Chinese, so on an English interface you are scanning the familiar Zotero commands for a Chinese string.

The repository top level shows a single file

The top-level listing for the repository holds exactly one entry, README.md. No package manifest, no plugin manifest and no source tree appear in that listing, which means the code you install is not code you can read from the repository root.

The build recipe is given in prose only: TypeScript and React, Webpack and ESLint, on top of the Zotero Plugin Template maintained by windingwind, with that author having granted permission for this use. How much of the compiled bundle in your Zotero profile is still template boilerplate, and how much is Suppr code, cannot be judged from what is published. Forking is a request to the maintainers rather than a task you can start locally.

Support badges say 6.0 and 9.0, the requirements list starts at 7

Compatibility is claimed in three places and the claims do not agree. One badge advertises Zotero 6.0 and up, a second claims Zotero 8.0 Compatible, and a third claims Zotero 9.0 Compatible. The requirements list under quick start is narrower, naming Zotero 7, Zotero 8 and Zotero 9 while skipping 6 entirely.

The troubleshooting entry then swings back down. If the translation option is missing after installation, the documented checks are a version of 6.0 or above, the plugin being enabled, a restart, and a right-click on the item again. A user on Zotero 6 therefore reads a requirements list that names no version for them, followed by advice inviting them to try anyway. Settle that before you install, by testing the Zotero build you actually run against the badges rather than against the requirements line.

Four manual steps stand between a profile and a first translation

Four manual steps stand between a fresh profile and a first translation. Open the GitHub Releases page, download the newest `.xpi` file, drag it onto the Zotero main window, then confirm the installation and restart Zotero. No shell command appears anywhere in the documentation, and no package name is published for a registry install.

For the first run, create a Suppr account, which the project describes as a WeChat QR code scan, then repeat the four steps above. Watch the attachment count on the item as it goes from one to two, because that new file is the entire output. Nothing further is documented beyond a settings panel for the language pair and a quality setting.

The walkthrough covers PDF, the FAQ promises five formats

The repository description names PDF, Word and PowerPoint. The FAQ lists PDF, PPT, WORD, EXCEL and HTML. The step-by-step walkthrough describes one type, a PDF document in the library, and leaves the other four to the web service pages linked from the project site.

That gap is easy to miss and costly in a large library. Attachments in Zotero carry their own metadata, and the documented flow drops a second file onto an item without saying how to rename it, tag it, or mark which of the two PDFs is the translation. Queue a folder of papers through the context menu and you finish with entries holding two attachments each and nothing to tell them apart. Cleaning that up later costs more attention than the translation did.

Automatic upload sits beside the privacy claim

Two feature bullets and one line of the documented flow sit awkwardly together. One bullet promises 隐私保护, glossed as the translation process being safe and reliable, and another says you never have to leave Zotero. Yet the flow states that the system uploads the document automatically and processes it remotely, and a network connection is listed as a system requirement.

Only the context menu runs on your machine. The FAQ on slow translation names three dependencies you cannot set: document size and page count, the quality of your connection, and server load. On weak wifi with a long scanned paper, your wait is a function of somebody else's queue. Unpublished drafts, air-gapped archives and offline reading all fall outside what this design can do.

Upgrades replace one file, and reuse terms are unstated

Upgrading is a file swap rather than a dependency update, because nothing is resolved at install time and nothing is resolved afterwards. The last push was on 2026-08-25, and the repository page carries no licence identifier, which leaves redistribution of the bundle unsettled for anyone planning more than a personal install. Put that question to the maintainers before you fork the code or repackage the `.xpi`.

The other recurring cost sits outside the repository. Quota, language pairs and quality settings live in the hosted service, so a plugin update does not change what an account is charged, and a service-side change can alter behaviour without any version bump in the `.xpi` file. Read the account terms on the site before your first upload.

Editorial conclusion

Adopt it if your library is PDF-heavy, you already trust a hosted translator with your papers, and a free character allowance covers casual reading. Stay away if you handle unpublished drafts, need offline work, or want to audit what runs, since the top-level listing shows no source and the bundle arrives prebuilt. Before installing, match your Zotero build against the 6.0 and 9.0 badges rather than the requirements line, and read the account terms at suppr.wilddata.cn before the first upload, because the quota and the language pairs you pick belong to the vendor, not to the `.xpi` on your disk.

Frequently asked questions

Official sources

1. Issues
2. Project website
3. README
4. Releases
5. WildDataX/suppr-zotero-plugin on GitHub