Privacy built around a temporary reading session

TXT and Markdown files are read locally. The original file is not uploaded to application storage.

When reading starts, the selected Blog article or custom text and an optional title are sent once to the configured AI provider to generate three questions.

The active session uses sessionStorage. Reader preferences and validated question cache entries use localStorage.

Generated questions may be cached for up to 24 hours. Raw content is not stored in that cache entry.

A signed HttpOnly cookie applies abuse limits. It does not contain article text, an IP address, or a browser fingerprint.

When configured in production, Google Analytics receives page paths plus fixed event buckets. It never receives reading text, titles, filenames, session IDs, questions, explanations, or answers.

The current release does not create an account or save completed sessions to a remote profile. Closing the tab ends the browser session data lifecycle.

The AI provider receives only the request needed to generate questions. Provider URL, model, instructions, and credentials are controlled by the server, not by the browser.

You can clear the current tab, remove LumenRead-owned browser storage, or stop before sending custom text for question generation.