Skip to content

Instantly share code, notes, and snippets.

@MaTriXy

MaTriXy/PRIVACY.md

Last active June 26, 2026 10:43
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save MaTriXy/e17d47f6bcc1085c6cd31d6aed6091b4 to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save MaTriXy/e17d47f6bcc1085c6cd31d6aed6091b4 to your computer and use it in GitHub Desktop.
Download ZIP
Xplain Privacy Policy
Raw
PRIVACY.md

Xplain Privacy Notes

Xplain is designed as an opt-in, local-first browser assistant.

Xplain does not sell user data, does not use user data for advertising, does not use analytics, and does not require an account.

What Xplain Reads

Xplain reads page content only after the user presses Read current page. It can capture:

  • page title
  • sanitized URL without query string or hash
  • domain
  • meta description
  • selected text
  • headings
  • up to 9,000 characters of visible main page text

The extractor skips hidden content, form fields, editable content, navigation, footers, asides, scripts, styles, SVG, and canvas content.

What Leaves The Device

Xplain downloads model, tokenizer, config, and weight files from Hugging Face. Hugging Face may receive standard network request metadata, such as IP address, user agent, and requested model artifact URLs, when model files are downloaded.

Page text, prompts, generated answers, and session state are processed locally by the browser runtime and are not sent by Xplain to Hugging Face, an Xplain server, or a third-party AI API.

Storage

In extension mode, Xplain stores the latest prompt, captured page context, and answer in chrome.storage.session so closing and reopening the side panel does not immediately lose work.

Model files are cached in browser storage by the local model runtime under a versioned xplain-gemma-4-e2b-* IndexedDB database. The in-app Clear model cache control clears the model chunk and metadata stores.

Chrome Web Store Privacy Disclosure Draft

Data handled locally:

  • Website content: page title, sanitized URL, domain, meta description, selected text, headings, and visible page text after the user clicks Read current page or Read selection.
  • User-provided content: prompts pasted or generated inside the side panel.
  • Generated content: local model output shown in the side panel.

Purpose:

  • Provide Xplain's single purpose: local AI explanations and visual briefings for the active page, selected text, or pasted context.

Transmission:

  • Model files are downloaded from Hugging Face.
  • Page content, prompts, and generated answers are not transmitted by Xplain.

Storage:

  • Session state is stored in chrome.storage.session.
  • Model artifacts are cached in IndexedDB.
  • Users can clear session state and model cache from the UI.

Remote Code

All JavaScript used by the extension is packaged with the extension. Hugging Face downloads are model data files, not extension JavaScript.

Public policy URL for Chrome Web Store submission: https://gist.github.com/MaTriXy/e17d47f6bcc1085c6cd31d6aed6091b4

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment