Privacy policy
Rookle Study is a mistake book for students 13 and older. This page explains what we store, why, who can see it, how long we keep it, and how you export or delete it.
1. What we store
Apart from what signing in needs, we keep only what studying needs:
- Mistakes: subject, topic, question text (transcribed from screenshots into text and formulas), your answer, the correct answer, the cause and your note on it.
- Review history: how many times you missed each question, whether each review was right or wrong, the next review date, and whether it is mastered.
- Topics: topics, chapters, prerequisites and sources (page, definition number) extracted from syllabi or lecture notes you give the assistant, and how well you know each one based on your reviews.
- Notes: the text you ask the assistant to put in your notes, its lengths (full, short, cheat sheet) and sources.
- Source: which app an entry came from (e.g. ChatGPT, Claude), when, and the short sentence you used to ask for it (e.g. “add this to my mistake book”).
- Sign-in identity: sign-in method (Google or GitHub), the user ID that service gives you, your verified email and display name, and when you confirmed “I am 13 or older”. We never see your password.
- Connections and settings: which assistants are connected, what they may do, when they were last used; your website settings; sign-in sessions (token hashes only) and an action log (who did what and when, without content).
- Save decisions: whether each question was saved or refused and whether it was later undone, as IDs and result codes only, no text.
We do not collect: the photos or screenshots themselves, full conversations, school, real name, phone number, address, location, contacts or payment details. Anything that looks like a password, key or contact detail is refused.
2. Screenshots
When you paste a screenshot into ChatGPT or Claude, that assistant transcribes it and sends us only the text; the image is never sent to Rookle. Screenshots you upload on our website or in the Rookle Study app are sent to OpenAI for transcription into text and formulas (Cloudflare Workers AI is used when OpenAI is unavailable). Each image is sent only for that one transcription (and re-sent once on a temporary error), the request tells OpenAI not to store it (store: false), and we do not keep the image either: it is cleared from memory once transcribed. When you ask the website’s AI to rewrite a note, that part of the note is sent to OpenAI the same way.
3. How we use it
- To save, find and review your mistakes, schedule reviews and show which topics you miss most.
- To let your connected assistant remind you of past mistakes when you ask about a topic.
- To protect the service: rate limits, abuse prevention and daily usage limits.
We do not build profiles, show ads, sell your data, train models on it, or analyze it together with other accounts. There are no social features: nobody else sees you and we never suggest strangers.
4. Who can see it
- You: everything above, in my mistake book.
- Assistants you connect (e.g. ChatGPT, Claude): can only read and add to your mistake book; you can revoke them under “Connections” at any time.
- The operator: only to fix problems, handle your requests or when the law requires, and only what is necessary.
Rookle Study and Rookle Memory (memory.rookle.dev) are separate services; accounts and data are not shared.
5. Service providers
- Cloudflare hosts the service and all data (Workers, Durable Objects storage, KV daily backups, Vectorize search); when OpenAI is unavailable, Workers AI transcribes screenshots and rewrites notes. Data may be processed and stored outside your country or region.
- OpenAI transcribes the screenshots you upload and rewrites notes when you ask (through the OpenAI API). Under OpenAI’s API policy this content is not used to train models; our requests ask OpenAI not to store it (
store: false), but OpenAI may keep request logs for up to 30 days for abuse monitoring. Processing may happen in the United States or elsewhere outside your country or region. - Google, GitHub: only to sign you in and confirm your email (Google:
openid email profile; GitHub:read:user user:email). - What your connected assistant (ChatGPT, Claude, …) sees in your conversation and screenshots is handled under that service’s own privacy policy.
6. How long we keep it
- Mistakes and review history and notes: until you delete them.
- Trash: deleted questions stay 30 days, then are permanently deleted.
- Daily backups: the last 7 days, for recovery; deleted together with your account data.
- Action log and save decisions: 180 days.
- Website sessions 7 days; connection access tokens 12 hours, refresh tokens 30 days.
7. Export and delete
- View: sign in to my mistake book.
- Export: “Export everything” (JSON or Markdown) under “Data”.
- Delete one: tell the assistant “undo that last one”, or delete it on the website (it goes to the trash first).
- Delete everything: “Delete account data” under “Data” removes all mistakes, review history and daily backups and revokes every connection. This cannot be undone.
- You can also email support@rookle.dev to see, correct or delete your data.
8. Age
Rookle Study is only for students 13 and older; connecting requires confirming “I am 13 or older”. If you are 13 to 17, please use it with your parent’s or guardian’s knowledge and consent. If we learn we have data from a child under 13, we delete it right away; parents can email support@rookle.dev at any time to ask for deletion.
9. Security
HTTPS everywhere; each account’s data lives in its own separate storage; connections use OAuth and you can revoke them at any time. If a security incident affects your data, we will tell you as soon as we can.
10. Changes and contact
If this policy changes in an important way, we will update the date on this page and show a notice on the website.
The service is run by Rookle placeholder, pending confirmation; privacy questions to support@rookle.dev.