Last updated: July 22, 2026
Tab-by: Tab Summarizer is a Chrome extension that extracts and compares structured information from open browser tabs. Local processing is the default. Optional cloud services can provide reusable extraction templates or infer a template when the browser's built-in model is unavailable.
When the user opens a Tab-by summary, the extension may process browser tab information such as tab titles, URLs, favicons, window and tab ordering, and page content visible in the browser. Page content may include text, headings, metadata, links, tables, and other page structure needed to create the summary and comparison views. This processing normally occurs on the user's device.
Tab summaries and user-generated extraction profiles are stored in Chrome session storage and are cleared when the browser session ends. An Analyze result may be kept in the current summary tab's session storage so it can be restored after a page refresh; it is invalidated when the corresponding redacted comparison input changes. A public extraction-profile catalog may be cached in Chrome local storage. Full HTML snapshots are requested from an open tab only when the user opens Snapshot; they are displayed on demand and are not stored by Tab-by.
Tab-by may download a static catalog of public extraction profiles from Firebase Hosting. Catalog requests do not include page content, tab URLs, or extracted values. The catalog contains reusable rules maintained by Tab-by.
For an unknown or partially supported page, Tab-by may request an existing public template after Google authorization. If local structure inference is unavailable, the extension asks before using cloud inference. These requests can include the website host, a generalized route shape, an abstract DOM skeleton, selector placeholders, value types, semantic hints, occurrence counts, and a structural fingerprint.
When the user clicks Analyze, Tab-by may send a bounded comparison payload containing field labels and selected extracted cell values. Before upload, URLs, email addresses, phone numbers, account handles, tokens, and identity-related fields are replaced. Chat text and free text from pages that cannot be treated as public are replaced in full with a placeholder. The service uses the remaining values only to correct Structured tables and interpret Similar patterns, and does not retain the comparison payload.
Cloud requests do not include raw HTML, real CSS selectors, complete URLs, query parameters, cookies, stored snapshots, or Chrome tab identifiers. Placeholder selectors and temporary row/page keys are mapped back to local data only on the user's device.
After a generated table is displayed, the user may choose to contribute its anonymized field rules to the shared template. Contributions are aggregated by site structure and are not used to create personal profiles. The service keeps bounded vote and consensus records and automatically expires low-use data.
Chrome Identity provides a Google OAuth token to authenticate cloud requests. The service does not persist the user's name, email address, OAuth token, or raw Google account identifier. One-way, short-lived hashes are used for rate limits, and a site-scoped one-way hash prevents duplicate contributions. Site-scoped hashes cannot be used to correlate votes across different websites. Authentication is not used to personalize template results.
Optional remote features use Google OAuth and Google Cloud services, including Firebase Hosting, Cloud Run, Firestore, and Gemini. Google processes data under its own Privacy Policy. Tab-by does not sell user data or share it with advertisers or data brokers, and does not use processed data for advertising or cross-site tracking.
The extension does not load or execute remote JavaScript. All extension code is packaged with the extension.
For privacy questions, contact the developer at: [email protected]