Last updated: August 9, 2026

ReadThread is designed to work without an account and without ReadThread-operated conversation storage.

Data stored locally

ReadThread stores the following inside the Chrome extension profile:

  • Provider configuration and the user’s chosen credential-storage mode.
  • API credentials in Chrome local or session storage.
  • Selected passages, bounded surrounding context, structured table/chart metadata, requested readable-page overview content, source URLs, page titles, conversations, and export metadata in IndexedDB.
  • For webpage charts, figures, ARIA grids, and complex tables, at most one bounded focused image may be stored with the conversation. The capture contains only the visible portion of the element and ReadThread does not scroll the page automatically.
  • For PDF conversations, exact user-selected text and zero, one, or two bounded capture images, depending on the capture method. Original PDF files and full-resolution pages are not retained.
  • Preferences such as output language and automatic-selection behavior.

Persistent extension storage is not a hardware-backed secret vault. Anyone who controls the device or browser profile may be able to access it. Session-only credential storage is available.

Data transmitted

ReadThread sends selected text and the displayed surrounding context, structured table/chart metadata when available, or the bounded readable-page content captured after Get page overview is chosen, together with source title and URL, conversation messages, model identifier, and the necessary API credential directly to the model provider configured by the user. The provider’s privacy policy and retention rules apply.

ReadThread does not transmit page content merely because a page is opened or text is selected. Transmission begins only after the user chooses an explanation, chooses Get page overview, or sends a follow-up message.

For PDF captures, ReadThread previews every image and names the selected provider before transmission. Exact-text captures can omit images for text-only providers. Images that are used are sent again with each follow-up because provider API calls do not retain prior visual input.

For a focused webpage visual, ReadThread preserves the exact selected text and structured data, clears Chrome’s selection tint, then previews the bounded image before transmission. Include visual context is enabled by default and can be turned off before the first request; disabling it keeps the image local. Providers without image support continue with exact and structured text only. At most one webpage image is sent with a request.

When the user chooses Continue in…, ReadThread places a bounded wrapper containing source material and conversation messages directly into the selected chat platform’s draft editor. The platform’s privacy and retention rules apply after the user reviews and sends the draft. ReadThread does not submit it automatically, read later replies, or handle platform authentication.

Website permissions

The toolbar action, keyboard shortcut, and selection context menu use temporary access to the active tab. The context menu receives text only after the user explicitly chooses Explain selection with ReadThread. The optional automatic-selection button requires website access explicitly enabled by the user. Disabling the feature removes that optional access.

Provider access is requested for the configured provider origin. Chat handoff requests access only to the selected supported platform. A pending handoff is kept in Chrome session storage for up to 30 minutes so it can resume after authentication, then it is filled once or discarded.

Accounts, analytics, and advertising

ReadThread has no ReadThread account, advertising SDK, analytics, tracking, or ReadThread-operated cloud conversation synchronization. ReadThread does not sell user data or use it for advertising, creditworthiness, or unrelated purposes.

Deletion and export

Users can export individual threads and delete individual threads or all local ReadThread data. Webpage and PDF image binaries are omitted from exports and external chat handoffs; source/capture metadata and an omission notice remain. Uninstalling the extension removes extension-local data according to Chrome’s behavior.

When Open precise reader needs to locate an exact PDF selection, the selected text is stored in Chrome session storage under a single-use random identifier. It expires after ten minutes, is removed when consumed, and is never placed in the reader URL. Expired records are ignored and pruned on the next handoff; Chrome also clears session storage when the browser session ends.

Changes and contact

Material policy changes will be published on this page and recorded in the open-source repository. For privacy questions, open a private-contact request through the project support page.