LatentLark Citation Bridge ("the extension") is a browser extension that captures the sources cited in AI research answers, checks them against public scholarly registries, and exports them as citations. It is built to collect as little as possible.

• No account, no analytics, no telemetry, no advertising, no tracking.
• We run no server. LatentLark never receives your data.
• What's on your screen is read only when you click Capture.
• To verify sources, source links and titles are sent to doi.org, CrossRef and DataCite.
• Zotero is optional. If you connect it, your chosen items go directly from your browser to Zotero.

When you click Capture on a supported page (Perplexity; Gemini and NotebookLM in beta), the extension reads, from that tab only:
- the links and titles of the sources cited in the answer (for NotebookLM, also the titles of the sources in your notebook's source panel),
- DOIs printed in the answer text,
- the question text, used only to name the export file.

It does not read pages you don't capture, your browsing history, form fields, passwords, or cookies.


| Recipient | What | Why |
|---|---|---|
| doi.org (DOI Foundation; resolves CrossRef and DataCite DOIs) | DOIs found in the source links or answer text | Check that each DOI exists and fetch its metadata |
| api.crossref.org (Crossref) | Source titles and DOIs, plus the developer contact latentlark@proton.me (Crossref's "polite pool" etiquette; it identifies the extension, not you) | Find matching scholarly records |
| api.datacite.org (DataCite) | Source titles and DOIs | Find matching records (datasets, preprints such as arXiv) |
| api.zotero.org (Zotero), only if you set it up | Your Zotero API key, user/group ID, and the citation records you choose to push | Create items in your own Zotero library |

These services get the request like any website you visit (including your IP address) and handle it under their own privacy policies: Crossref, DataCite, DOI Foundation, Zotero. The extension sends them nothing else.


Everything is kept in your browser's extension storage (browser.storage.local) on your device:
- the last capture and its verification results,
- a daily capture counter (free tier),
- the "Pro" switch (in this version a local setting, with no purchase or licence check),
- the de-duplicated library of past captures (Pro),
- Zotero settings: library type, user/group ID, optional collection key and API key. The API key is never shown again in the popup, never written to logs, and is sent only to api.zotero.org.

Nothing is synced to a LatentLark server; there isn't one. To delete it all, use Settings & Pro → Library → Clear and Zotero → Remove, or remove the add-on in about:addons, which deletes its storage.


This version handles no payments and collects no payment information. If paid features are added later, this policy will be updated before that release.


The extension is a research tool. It is not directed at children and knowingly collects no data from anyone.


Firefox shows what the add-on transmits before you install it, and lists it under about:addons → LatentLark Citation Bridge → Permissions and data:
- Required: website content. The source links and titles (and DOIs printed in the answer) from the page you capture are sent to doi.org, CrossRef and DataCite for verification. Nothing is sent until you click Capture.
- Optional: authentication information. Only if you set up Zotero: Firefox asks before your own Zotero API key is used with api.zotero.org. You can withdraw it in about:addons at any time; Zotero push then stops working.

Site access (perplexity.ai, gemini.google.com, notebooklm.google.com and the registries) can also be turned off per site in about:addons. The popup then shows an "Allow access" button.

Data is used only to provide the add-on's single purpose. It is not sold, not used for advertising, and not read by humans.


We will update this page and the version/date above when the extension's data handling changes.


latentlark@proton.me **