On this page
About
Privacy
Last updated
The German Datenschutzerklärung is the binding version. This page translates it under Controller, legal bases and your rights and lists every stored field in detail.
Short version: signed out, the editor never sends your diagrams anywhere unless you make a short link; signed in, it keeps them in your account (private, and you can turn that off). Payments for Pro go through Polar, and mock interviews and Design partner chats are sent to an AI provider (Anthropic) only when you start one. There are no analytics or tracking scripts, no cookies for counting and nothing that follows you; the site only counts, per day, how often a few things happen (say, "the editor was opened 412 times on 6 October"), with nothing about who. An account is optional (for practice, short links and keeping your diagrams across devices), and an email address is kept only if you sign in with one or ask for email reminders. The rest of this page lists exactly what is kept, where, and for how long.
Controller, legal bases and your rights #
Controller. The person named in the Impressum, which also gives the contact address for every data protection request.
| Purpose | Data | Legal basis | Kept for |
|---|---|---|---|
| Providing the website | IP address (rate limiting only), request logs with path, status and, signed in, user id | Art. 6(1)(f) GDPR (secure operation) | Cloudflare Workers Logs' retention period |
| Account and sign-in | Provider id, display name, email address, password hash, sessions | Art. 6(1)(b) GDPR | Until the account is deleted; sessions 30 days, renewed with use |
| Progress, diagrams, short links | What you save in your account | Art. 6(1)(b) GDPR | Until you or the account's deletion removes them |
| Paid plans and credits | Polar customer and subscription ids, plan, status, credit ledger | Art. 6(1)(b) GDPR | Until the account is deleted; Polar keeps its statutory records |
| AI features (mock interviews, Design partner) | Your input, your design, transcripts and verdicts | Art. 6(1)(b) GDPR | Transcripts and chats 90 days, verdicts until the account is deleted |
| Contact form on the Impressum | Name (optional), email address, message, IP address (rate limiting and Turnstile only) | Art. 6(1)(f) GDPR (answering enquiries), Art. 6(1)(b) GDPR where it concerns a contract | Not stored on the server; the email stays in the mailbox until the enquiry is settled |
| Email reminders and product news | Email address, settings, time zone | Art. 6(1)(a) GDPR (reminders), Art. 6(1)(f) GDPR and § 7(3) UWG (product news) | Until you unsubscribe or delete the account |
| Preventing abuse of sign-up credits | One-way hashes of sign-in identity and IP address | Art. 6(1)(f) GDPR | IP hash 7 days, identity hash permanently |
| Operation and security | Audit log, error events | Art. 6(1)(f) GDPR | Audit log 400 days, events 30 days |
| Usage counts | Daily counters with no personal reference | No personal data | 400 days |
Account data is needed to use an account. The editor, docs and practice work without one.
Cookies and browser storage. Only technically necessary cookies are set (see Cookies below), under § 25(2) no. 2 TDDDG.
Recipients. Cloudflare (hosting, database, storage, email delivery, AI gateway, speech recognition, Turnstile); Polar as merchant of record for payments, invoices and VAT; Anthropic (the Claude model), only when you start an AI feature; GitHub and Google, only if you sign in with them.
Transfers outside the EU. Some recipients are based in the USA or process data there. Transfers rely on an EU adequacy decision (Art. 45 GDPR) where the recipient is certified, and otherwise on standard contractual clauses (Art. 46(2)(c) GDPR).
Your rights. You have the right of access (Art. 15), rectification (Art. 16), erasure (Art. 17), restriction (Art. 18), data portability (Art. 20) and to object to processing based on Art. 6(1)(f) GDPR (Art. 21). You can withdraw consent at any time with effect for the future (Art. 7(3)). You can also complain to a data protection supervisory authority, in particular where you live (Art. 77 GDPR).
Automated decisions. No decision with legal effect under Art. 22 GDPR is made automatically. The AI interviews' verdicts are practice feedback.
In your browser #
The editor, the docs and practice work without an account and without a server. What they remember stays in your browser's storage on this device:
| Key | What it holds |
|---|---|
proschi.docs |
Your diagrams in the editor: their text, file names and which one is open |
proschi.docs.sync |
Cloud sync, signed in: which account it syncs with, whether you turned it off, per diagram the version your account has and a fingerprint of its content then (to tell what changed since), and which diagrams in this browser you have not (yet) agreed to add to the account |
proschi.playground.source |
A diagram saved by an older version of the editor, read once to carry it over |
proschi.playground.files |
Whether the editor's list of diagrams is shown or hidden |
proschi.playground.collapsed |
The names of the folders you closed in the editor's list of diagrams |
proschi.practice |
Practice progress: per problem, your latest design, how many times you ran the tests and whether you solved it |
proschi.practice.runs |
Per problem: how many test runs, the runs it took to solve it, and the monthly cost of your cheapest solving design and of the reference solution, for the badges on a copy of the site without accounts |
proschi.practice.last |
The last practice problem you opened, for "Continue" on the practice page |
proschi.practice.mistakes |
The known mistake your last failed test run matched, per problem (its name, title and when), for "Your next steps" on the practice page; signed in, joined by the copy your account keeps |
proschi.practice.plan |
The interview-date plan (Max): your interview date, the minutes a day, the day you set it, the last day's list and what you ticked by hand; signed in, the date and minutes are a copy of what your account keeps |
proschi.practice.goal |
What you are on the practice page for (solving problems, interview prep, a daily habit), whether you answered or skipped that question, and whether you switched the daily streak on or off; signed in, a copy of what your account keeps |
proschi.interview.<problem> |
Scripted interview mode in progress on a problem: the length, the phase, the time spent per phase, the questions asked, your estimates, the last test run and the self-review ticks; removed when you leave interview mode |
proschi.interview.history |
The summaries of your last 50 finished scripted interviews (time per phase, questions, estimates, tests, self-review) |
proschi.aiInterview.<problem> |
Signed in: the id of the mock interview open on a problem, so a reload resumes it; the interview itself is on the server, below |
proschi.prep.askTip |
That you used or dismissed the tip that selecting text on an interview prep step asks the Interview coach |
proschi.aiInterview.consent |
That you agreed once to what a mock interview sends to the AI provider |
proschi.achievements |
Badges earned on a copy of the site without accounts: which, when, and when the page showed them to you |
proschi.cards |
Daily review, on a copy of the site without accounts: every card review (which card, your rating, when, how long it took and your local date) |
proschi.cards.outbox |
Daily review, signed in: reviews not yet sent to the server, kept until it answers |
proschi.solves |
On a copy of the site without accounts: the local date you first solved each problem, for the daily streak and its badges |
proschi.challenge |
The daily challenge, on a copy of the site without accounts: each day's result (your answers, how long each took, the points, and your local date, for the daily streak) |
proschi.challenge.guest |
The daily challenge, signed out: the last one you played (answers, times, points and the card reviews it made), so it can be saved to your account when you sign in |
proschi.challenge.progress |
The daily challenge you are in the middle of: the day, your answers so far and how long each took, so a reload carries on where you were; removed when you finish |
proschi.game.meta |
Scale or Fail, signed out: your progress (Blueprints, unlocks, perks, the furthest wave and highest difficulty per scenario, what you have seen) |
proschi.game.outbox |
Scale or Fail, signed out: your finished runs (their moves), purchases and equipped perks, sent to your account when you sign in and then removed |
proschi.game.run |
Scale or Fail: the run you are in the middle of (its setup and your moves so far), so a reload carries on; removed when it ends |
proschi.game.daily |
Scale or Fail, signed out: the day and score of the last daily run you played |
proschi.game.settings |
Scale or Fail: sound on or off, and the run's speed |
proschi.game.days |
Scale or Fail, on a copy of the site without accounts: how many runs you finished on each local date (the last 400 days), for the daily streak |
proschi.goal |
On a copy of the site without accounts: your daily goal, in cards a day |
proschi.recap |
The week of the last weekly recap you dismissed, so it is not shown again |
proschi.onboarding |
Which first-run tours you have seen, the Arcade's first-wave tutorial and its intro to the twists included (in session storage when local storage is blocked) |
proschi.theme |
Light, dark, or following your system |
proschi.chunkReloadAt |
Session storage only: when the page last reloaded itself after a site update, so it never loops |
proschi.metrics.once |
Which usage counts this browser has already sent where one counts only once (below): in local storage for the first edit and a first solve signed out (with the problem's id), in session storage for the rest. Never sent anywhere |
Clearing the site's data in your browser removes all of it. Apart from the theme, the docs pages and their live examples store nothing.
Share links carry the whole diagram in the part of the address after
#, which browsers never send to a server. Whoever you give a link to can
read the diagram; nobody else sees it. The same goes for an embed made
signed out (/embed/#code=…).
Short links (/s/<id>, signed in only) are different: the diagram is
stored on the server, below.
Share buttons (a daily run, a solve, a badge, a profile) only hand text to your device's share sheet or clipboard; nothing is sent to us or to anyone else until you paste it somewhere. A badge's share text links to your public profile, so it is offered only while your profile is public.
No tracking #
No analytics, advertising or tracking scripts, no third-party cookies, and no requests to other sites: fonts and scripts are served by proschi.app itself, and every page's Content-Security-Policy forbids loading them from anywhere else.
Usage counts #
To know whether the site is useful (do people who open the editor go on to edit? do people who start a problem solve it?), proschi.app counts how many times a day each of these happens:
| Count | When |
|---|---|
landing_view |
The home page is opened (once per browser tab session) |
editor_open |
The editor is opened (once per session) |
editor_first_edit |
The first edit in the editor (once per browser, ever) |
simulation_run |
The editor's Results or HLD tab or the load overlay is first shown (once per session) |
test_run |
The tests of a practice problem are run, or the editor's Results tab is first shown in a session |
share_link_created |
A share link is copied |
export |
A diagram is downloaded (image, .proschi file, backup) or copied as Mermaid |
practice_open |
The practice page is opened (once per session) |
problem_start |
A problem is opened (once per problem and session) |
problem_solve |
A problem is solved for the first time: signed in, counted by the server once it verified the solve; signed out, sent by the page (once per problem and browser) |
card_review_session |
A daily review session is finished |
challenge_complete |
A daily challenge is finished |
arcade_run_start, arcade_run_end |
A Scale or Fail run starts, ends |
sign_in |
Someone signs in (counted by the server) |
email_opt_in |
Someone asks for email reminders on the account page (a confirmation link is sent) |
email_reminder_sent |
Reminder emails sent (counted by the server, how many each hour's run sent) |
What is sent and kept: the page sends only the names of these events
(POST /api/metrics), a few at a time. The server adds them to one number
per day (UTC) and event, e.g. 2026-10-06, editor_open, 412, and keeps
nothing else: no user id even when you are signed in, no IP address, no
cookie or device id, no page address or referrer, no problem id, no time of
day. Your IP address is used only for the minute-long rate limit that keeps
the counts from being flooded, and is never written down. Because a count
cannot be traced back to anyone, it is not part of Download my data and
stays when you delete your account. The daily counts are kept for 400 days,
then deleted.
Turning it off: the page sends nothing when your browser sends Do Not Track or Global Privacy Control. Copies of the site without accounts (such as a local build) never send anything.
With a practice account #
On proschi.app you can sign in with GitHub, Google or an email address and password to keep practice progress and your diagrams across devices, see how others did and make short links to diagrams. Without signing in, nothing about you reaches the server (the usage counts above are about events, not about you).
What is stored
-
Your account: the sign-in provider and the user id it gives Proschi, a display name (your GitHub login or Google first name, which you can change), whether you chose to appear on the leaderboard, when the account was created, and its plan (free or Pro; it sets how much the account may keep and how many mock interviews it gets).
-
Sessions: a hash of each session's cookie, with when it started and when it expires. A session lasts 30 days and renews while you use it; expired ones are deleted daily.
-
If you sign in to a Proschi app on a phone, or to the
proschicommand line tool: a hash of each of its tokens (an access token that lasts an hour and a refresh token that lasts 60 days), with whether it is the site's cookie or which app token it is, when it was issued, when it expires and, for a refresh token, when it was exchanged for new ones, and what the tokens may be used for (their scopes: for the command line tool, reading your name and your synced diagrams; never billing, credits, the AI features or account changes). Settings lists these sign-ins with their scopes. Rotated refresh tokens are kept until they expire, to notice a stolen copy being used. During sign-in a hash of a one-time code, which the app exchanges for its tokens within 60 seconds, is kept with the app's address to return to (for the command line tool, the localhttp://127.0.0.1:<port>/callbackaddress it listens on). Expired ones are deleted daily. -
Practice progress, per problem: how many times you ran the tests, your latest design, when you first ran and last ran it, when you solved it and in how many runs and on which local date, the cost and p99 of your best solving designs (as the server measured them) and when each was reached, and the simulation and problem versions it was recorded under. When you sign in, designs in your browser that your account lacks are uploaded to it, and the server re-runs their tests before it counts a solve.
-
Known mistakes: when a failed test run on a practice problem matches a common mistake, the problem, the mistake's name and title, and when it last matched (the newest 100 per account; one row per problem and mistake). They are kept so "Your next steps" can suggest retrying the problem on any device you sign in on. Mistakes kept in your browser are sent when you sign in; signed out, they stay in your browser. Free on every plan.
-
Daily review, per card review: which card (and its version), your rating (again, hard, good or easy), when, how long you took to answer and your local date; and per card the schedule computed from them (when it is due next and how well you know it). Reviews made signed out stay in the page and are gone when you close it.
-
Your daily goal (cards a day), and what you are on the practice page for (solving problems, interview prep, a daily habit; whether you answered or skipped that question, and whether you switched the daily streak on or off). Signed in, that answer is kept with your account instead of only in this browser: the first time you sign in, the browser's answer is copied to it if the account has none. The daily streak and the weekly recap are computed from your card reviews, solve dates, daily challenges and game runs (by the local date each was sent on, kept with them); nothing more is stored for them.
-
The daily challenge, per day you play it: when you saw its first card, your answers to its five cards, how long each took, which were right, the points, the score, when you sent it and your local date then. Your rank is computed from everyone's scores each time it is shown. A challenge played signed out stays in your browser until you sign in, when it is saved to your account.
-
Scale or Fail, the system design game: your progress (Blueprints, unlocks, perks, the furthest wave and highest difficulty per scenario, the cards and incidents you have seen) and every run you submit: its scenario, seed, loadout, the moves you made, the score, when you played and your local date then. Runs played signed out stay in your browser and are sent when you sign in.
-
Your diagrams in the editor, while cloud sync is on (it is, once you sign in, until you turn it off in the editor's File menu): each one's file name, text and the imported files a share link brought, a version number and when it last changed. The diagrams already in your browser are uploaded when you first sign in. They are private: they never appear on a profile or a leaderboard, and only you can read them. Deleting a diagram leaves a marker (its id, version and when) for 30 days, so your other devices delete their copy too; its text is removed at once. Your account keeps at most 5 diagrams on the free plan, 64 KiB each (Pro: no practical limit, 512 KiB each); the others stay in your browser only, marked "This browser only" in the editor's list of diagrams, and are never lost. Move to this browser only, in a diagram's actions there, takes it out of your account (and your other devices) to free a slot. Turning sync off keeps the copies in your account until you choose Delete my cloud copies in the File menu or in Settings.
Shared computers. The browser remembers which account its diagrams were synced with. When a different account signs in, the diagrams synced with the previous one are removed from this browser (they are safe in that account) and nothing of it is uploaded to the new one. Other diagrams in the browser (never synced, or with edits the previous account never received) stay, and are added to the new account only if you answer yes to "Add N diagrams from this browser to your account?" (at most as many as your account has room for). Signing out gives you the choice: Sign out keeps the synced diagrams in this browser (the default, as before), Sign out and remove synced diagrams from this browser removes them (edits not yet saved to your account stay).
-
Versions of your synced diagrams, each one you tag (from the editor or the command line tool): a copy of the diagram's file name, text and imported files as your account had them then, the tag you chose (such as
2026-11-01), an optional note, when you made it, its size and a fingerprint (SHA-256) of its content. A version never changes. Like your diagrams they are private: only you can read them, andcloud:references to them resolve only in your own account. Up to 20 per diagram on the free plan. Deleting a version removes its copy, note and fingerprint at once and keeps only its tag, number and dates, so the tag isn't reused; deleting the diagram (or every cloud copy, or the account) deletes all its versions, tags included, at once. -
Short links, each one you make with Share → Short link with preview (or Share → Embed, signed in): the diagram's text and the files it imports, its title, a PNG preview of the diagram that your browser rendered (none if it could not), and when you made it. Anyone with the link can open the diagram and its preview, without an account; the link's id is ten random characters, so it can't be guessed, but treat it like the diagram itself. Your name is not shown with it. Share → Your short links lists and deletes them; a deleted link stops working at once, though chat apps and link previews that already fetched the preview may keep their copy.
-
Achievements: which badges you earned, when, and when the page first showed them to you. The skill map and the badges are computed from the reviews and progress above each time you open them; nothing else is collected for them.
-
Only if you ask for email reminders: your email address and its settings (Email reminders, below).
-
Only if you make an interview-date plan on Max: the interview date, the minutes a day, the day you set it, and today's list as your browser last made it (each item's title and whether it is done), for the evening reminder. Removing the plan on the roadmap deletes it, and it is in Download my data.
-
Only if you sign in with an email address: that address, a hash of your password, and when you confirmed the address (Email sign-in, below).
-
If you sign in with GitHub or Google: the verified email address they give, as a contact address for account and product news email only (Product news, below).
-
The last day (UTC date only, no time) you used your account, for counting how many people are active (The site's operator, below).
What is not stored: your avatar, and the access
tokens GitHub or Google hand over during sign-in (they are used once, to
read your user id, name and verified email address, and dropped). Proschi
asks GitHub for your public profile and read-only access to your email
addresses (user:email), and Google for openid profile email. The
verified address is used for the free sign-up credits' checks
(only a one-way keyed hash of it is kept for those) and, since November
2026, kept as your contact address (Product news). It is
never a way to sign in or to find your account.
Cookies: __Host-proschi_session keeps you signed in (HttpOnly, sent
only to proschi.app). proschi_oauth lives for ten minutes during sign-in to
tie the provider's answer to your browser. Neither is used for anything else.
An app sends its access token with each request instead of a cookie; when
it signs in through your browser, no session cookie is set there.
Cookies of the operator: __Host-proschi_admin signs the site's
operator in to the admin panel; visitors never get it.
Logs: each request to the server is logged with a request id, its method, path (never the query), status, duration and, when signed in, your user id. The logs live in Cloudflare Workers Logs and are kept for Cloudflare's retention period.
What others see: practice statistics are aggregates over everyone, such
as how many solved a problem and how your design's runs, cost and p99 compare.
Your display name and number of solved problems appear on the leaderboard,
your display name and the monthly cost and p99 of your best solving designs
on each problem's leaderboards,
your display name and daily challenge score on that day's challenge
leaderboard, and your display name and best game score on the game's
leaderboards, only if you opt in ("Show me on the leaderboard", in Settings
at #/me). Without it, your challenge score and your best designs still count
toward the number of players and everyone's ranks, but your name is never
shown.
Your public profile: opting in also makes a public profile, linked from
your name on either leaderboard, at an address with your user id
(proschi.app/u/<your id>). That address can be shared and found by search
engines, and its link preview (the page's title, description and a picture
card) shows your display name, how many problems you solved, your badge
count, your current streak and your interview-ready score, all from the
list below. The profile shows exactly this, and nothing more:
- your display name, and the month you joined;
- the problems you solved (which ones and their difficulty, never your designs);
- your current and longest daily streak, in days;
- your current and longest daily challenge streak, in days, and your best daily challenge score;
- your interview-ready score and your mastery of each topic, in whole percent;
- the badges you earned, with the day each was earned.
Your designs, your daily goal, your card reviews and when you made them, how many cards you reviewed, your daily challenge answers, your test runs and their costs, your sign-in providers and your sessions are never on it. Turning the option off hides the profile at once: its address then answers "not found", the same as an address no user has, so nobody can tell whether you have an account. Its picture card stops being served at once too, but link previews that a chat app or social network already fetched are kept by them, beyond our control.
Your mastery certificate (Max): turning it on in Settings › Exports &
certificate makes a public page at a random address
(proschi.app/c/<certificate id>, not your user id), with a link preview,
that shows your display name, your level and XP, your interview-ready score,
your mastery of each topic in whole percent, the problems you solved (which
ones, never your designs), your longest streak and the day it was first
turned on. It is read from your progress each time it is opened. Turning it
off, or a plan without it, makes the address answer "not found" at once; it
comes back at the same address when you turn it on again.
Plans, credits and payments #
Pro, Max and top-ups are paid through Polar, the merchant of record: Polar runs the checkout and the billing portal on its own pages, charges your card, handles VAT and sends the invoices, under its privacy policy. Proschi never sees or stores your card, billing address or the email address you give Polar.
What Polar gets from Proschi: your account's id (so a payment lands on the right account) and which product you chose. Nothing else is sent.
What is stored here, once you buy something:
- your Polar customer id and subscription id, the subscription's status (active, canceled and so on), which plan (Pro or Max, monthly or yearly), when the paid period ends, whether it is canceled at that date, and when this was last updated;
- your credits, as a ledger: the sign-up credits, each month's credits, each top-up bought (with Polar's order id), and each charge with what it was for (a mock interview, Design partner, a verdict, your next steps), its id, the credits and whether it ran in Deep mode, with the time of each;
- for every message Polar's webhook sends: its id, its type and when it arrived, to apply it only once. It holds nothing about you.
- a log of those messages for the site's operator, kept 90 days: when each arrived, its id and type, what was done with it and the account it was about; never its content.
- if the operator gives you Pro for a while, or adds or takes back credits (say, after a support request), the date the free Pro ends and those credits in the ledger; the reason is kept in the operator's audit log.
What it is used for: only to know your plan, renewal date and how many credits you have left. It is shown in Plan & billing on your account page, with your recent use.
How long: as long as the account. Deleting the account cancels a subscription still running at Polar at once, then removes all of the above (the webhook log keeps its rows without your account id until they are 90 days old; the list of applied message ids names no account). Polar keeps its own records of the payments, such as invoices, as the law requires of it.
Live share links (Pro) #
On Pro you can make a live share link to one of your synced diagrams. What is stored: a random token (the link's address), which diagram it opens, and when it was made and, once you revoke it, when. Anyone with the link sees the diagram as it is now, read-only, while you are on Pro; nothing about who opens it is kept. A revoked link stays as that row so its token is never reused. The links are in Download my data, and are deleted with the diagram or the account.
Mock interviews #
A mock interview (paid for in credits, on any plan) is started only by you, knowing it is sent to an AI provider: the problem, your diagram and everything you write or dictate in the interview are sent to Anthropic's Claude, through Cloudflare's AI Gateway, to ask the questions and write the hints and the verdict. The provider uses them only to answer, not to train its models.
What is stored: the interview's transcript, kept 90 days and then deleted, and its verdict summary (pass or fail, and what went well and wrong per skill), kept for your interview history. Both are private, part of Download my data, and deleted with your account. A daily job deletes each transcript, and the design sent with it, once the interview is 90 days old; the verdict stays. On Pro, your next steps are picked after each verdict (below). Dictation, where your browser offers it, is done by the browser; Proschi receives only the text.
Design partner #
Signed in, Design partner (paid for in credits, on any plan) is a chat about your design, in the editor or on a practice problem, started only by you. Each message you send goes to the server with your design's source (in the editor, also the files it imports; on a practice problem, the problem's id). The server runs the simulation and the tests on it and sends your message, the design, those results, the problem and the chat's earlier messages to Anthropic's Claude, through Cloudflare's AI Gateway, to write the answer and any changes it proposes. The provider uses them only to answer, not to train its models. On a roadmap problem's interview prep steps the chat is the Interview coach: each message also sends the step you are on and your notes of the steps (your clarifying questions, the estimate worksheet's rows and their checks, your wrap-up answer), which go to the provider the same way and are not stored.
What is stored: each chat (where it was opened, whether it is the Interview coach, and on a practice problem which one), its messages and the partner's answers, the changes it proposed with the server's check of each and whether you accepted or rejected it. A chat is private, part of Download my data, and deleted with your account; a daily job deletes it 90 days after its last message. What each message cost is in the credits ledger (above).
Your next steps (Pro) #
On Pro, after a mock interview's verdict, Proschi asks Anthropic's Claude to pick your next 3 steps. It sends the verdict, your weakest card topics, which roadmap problems you have solved and your mastery of each card topic. The provider uses them only to answer, not to train its models.
What is stored: each set of picks (the problem or card topic ids, the reason for each, when, and the interview that prompted it; older picks made after a design review, a feature Design partner replaced, keep the reviewed problem instead). They are private, part of Download my data, and deleted with your account.
Feedback #
Signed in, Send feedback (in the help menu, the footer and Settings → Contact us) sends a message to the operator: a bug, a question, a proposal or anything else. Only you start it, and the dialog lists what goes with it before you send.
What is stored: the type, the subject and message as you wrote them, your account (so the operator knows who wrote and can reply), and the context the dialog shows: the page's address without its query, your browser's user agent, the window size, your plan and the site's version. If you tick Attach my current diagram, its source; if you add screenshots (up to 3 PNG, JPEG or WebP images of 5 MB), those, stored privately in Cloudflare R2 where only the operator's admin panel can open them. Nothing is emailed and nothing is shared. The operator can mark a message new, in progress or done and keep an internal note on it.
The operator answers, when one is needed, at your account's confirmed email address (your email sign-in, the address GitHub or Google gave, or your email reminders' address), if it has one. Your messages are in Download my data (the screenshots by type and size, not the operator's notes) and are deleted with your account, screenshots included; the operator can also delete one. You can send up to 5 messages an hour and 20 a day.
The site's operator #
Proschi has one operator, who signs in to an admin panel (/admin/) with a
passkey. It reads the database directly, so the operator can see what the
server stores about an account (listed above), with these limits: the email
reminders' address is shown masked (a***@example.com) except on a
message you sent with Send feedback, where your confirmed address is shown
so the operator can reply, and session tokens
exist only as hashes. The panel shows aggregates (accounts, sign-ups, the
usage counts) and, per account, the display name, sign-in providers and
their user ids, when it joined and was last active (a date), its progress
and counts (diagrams and versions too), its plan, and its short links, but
not your diagrams', versions' or designs' text.
The operator can block an account that breaks the rules (spam in short links, an abusive name): it is signed out everywhere, cannot sign in again (the page says the account is blocked), its profile is made private and its short links stop working while it is blocked. The operator can also rename an account, make its profile private, sign it out, remove its email address, delete a short link, or delete the account as Delete account does. To ask about a block, open a GitHub issue.
What the panel keeps of its own:
- An audit log of the operator's actions: when, which action (for example "block", with the reason given), and the id of the account or short link it was taken on; never a name or address. Kept for 400 days, then deleted. It stays after an account is deleted, holding only that account's id.
- App events for the server's health, kept 30 days: server errors with
the route and request id, rate limits hit, failed sign-ins, sign-ins of
blocked accounts, failed reminder emails and the scheduled jobs' runs.
They never hold a user id, an IP address, a query or an email address;
ids in a route are replaced by
:id. - The last active day of each account (above): a date, written at most once a day, deleted with the account and part of Download my data.
Email sign-in #
You can sign up with an email address and a password instead of GitHub or Google. What is kept, besides the account above:
- the address (lowercased), and a normalized form of it the checks below
compare (dots and
+tagdropped for Gmail,+tagfor other domains); - a hash of your password (PBKDF2-SHA256, 100,000 rounds, a random salt of its own). The password itself is never stored or logged;
- when you confirmed the address;
- the confirmation and password reset links, as hashes, until a day or two after they expire (a confirmation link works for 24 hours, a reset link for an hour, each once). A reset signs you out everywhere else.
- failed sign-ins, to slow down password guessing: a count per address and per IP address, each stored as a one-way keyed hash (never the address or the IP), forgotten after a day.
The emails (the confirmation, the password reset, and a note when someone
tries to sign up with an address that already has an account) come from
reminders@proschi.app through Cloudflare Email Service. Sign-up, sign-in
and password reset use Cloudflare Turnstile
to tell people from bots: it runs in your browser, and Proschi only checks
its yes or no with Cloudflare.
Unlinking the email sign-in in Settings (when the account has another sign-in) stops it working; deleting the account deletes all of it.
Free sign-up credits #
A new account gets free credits once it is confirmed (an email account after its confirmation link, a GitHub or Google account at once), at most once per person. So that throwaway accounts can't collect them, the server keeps:
- for each sign-in identity it has checked (GitHub or Google and its user id, and your normalized email address, also the verified one GitHub or Google gives), a one-way keyed hash, which stays after the account is deleted, so the same person doesn't get them twice;
- a one-way keyed hash of the IP address an account signed up from, with the time, for "at most 3 accounts a week from one IP"; deleted after 7 days. The IP itself is never stored;
- for each account, whether it got the credits and why not (for example a disposable email domain).
An account that doesn't get them works as usual and can buy credits.
Email reminders #
Signed in, you can ask for reminders by email on your account page
(#/me). It is off until you turn it on, and nothing is emailed until you
follow the confirmation link sent to the address (it works for two days).
What is stored, in one row per account, and nothing else:
- your email address, and when you confirmed it (empty until you do) and when the last confirmation email was sent;
- your browser's time zone (such as
Europe/Berlin), read when you enter the address, so reminders arrive in your evening and the recap on Monday morning; - which reminders you want: streak at risk, cards due, weekly recap (an interview-date plan on Max is named in the evening ones);
- a random secret for the unsubscribe link in each email;
- the last reminder sent: your local date, the time and which kind (for the one-a-day limit), how many were sent in a row with no practice in between, and when reminders were paused for that reason;
- when the row was made and last changed.
What it is used for: only to send these reminders, at most one a day: your streak would end today and today's goal isn't met (in your evening), 5 or more review cards are due (in your evening, when no streak reminder was sent), what is left on your interview-date plan's list today (Max, in your evening, when neither of those was sent; they name it too), and last week's recap (Monday morning). Whether one is due is worked out from the practice data above, at the time. After three reminders in a row with no practice in between, they pause (the third email says so) until you resume them on your account page. Your address is never shown to anyone, never used for anything else and never given to anyone else. The emails are sent through Cloudflare Email Service, which handles the address only to deliver them; they carry no tracking pixels and no tracked links.
How long, and how to remove it: until you remove it. Every email has an unsubscribe link (also offered by your mail app as a one-click "Unsubscribe"), which deletes the row at once; so do Remove address on your account page and deleting your account. Changing the address replaces the old one, which then needs confirming again. Log lines of the hourly reminder run hold your user id, never the address, when a send fails.
Product news #
Now and then (a few times a year) Proschi emails account holders about new features, and may show the same news once on the practice page, sometimes with a gift of credits. The operator writes and sends these from the admin panel; every send is in its audit log with a reason.
Who gets the email: accounts with a confirmed email sign-in address, or the verified address GitHub or Google gave at your latest sign-in with them, unless you turned product news off. The address you gave for email reminders is never used for it. Emails about your account (confirmations, password resets, receipts) are not product news and are not affected.
What is stored:
- the contact address from GitHub or Google (one per account, replaced at each sign-in with them, with which provider gave it and when);
- whether you turned product news off, when, and how (in Settings, or with the link in an email);
- per campaign that reached your account: whether its credits were given, whether the email was sent, failed or skipped (and why: turned off, or no confirmed address), and when the announcement was shown, dismissed or its button clicked.
How to stop it: switch off Product news in Settings › Notifications, or use Stop product news at the bottom of any of these emails (also offered by your mail app as a one-click "Unsubscribe"; it works signed out). Once off, no product news email is sent to you. The announcement on the practice page goes away when you dismiss it. All of it is part of Download my data and is deleted with your account. The emails are sent through Cloudflare Email Service and carry no tracking pixels or tracked links.
Your controls #
In Settings on the practice page (#/me, from the account menu):
- Download my data saves everything stored about you as JSON, your synced diagrams and their versions included, your live share links, your plan (and any free Pro the operator gave), subscription, Polar webhook log entries, interview credits, mock interviews with the picks for your next steps, Design partner chats, the messages you sent with Send feedback, every session with its kind (the site or an app) but never the token hashes; short links come with their diagrams and the address of each preview, email reminders with their address and settings (the unsubscribe secret stays out), and product news: whether it is on, the contact address, and the campaigns that reached you.
- Delete my cloud copies, in Settings or in the editor's File menu once cloud sync is off, removes every diagram from your account, their versions with them, and turns cloud sync off in that browser; those in your browser stay. A single version can be deleted on its own.
- Delete account removes your account, sessions (apps' too), progress, card reviews, daily challenge results, badges, game progress, short links, synced diagrams and their versions and live share links, the email reminders' address, the contact address and product news settings, the billing data, interview credits, mock interviews and the picks for your next steps, and your feedback messages with their screenshots from the server at once (a Pro subscription is canceled at Polar first); the progress and diagrams in your browser stay. Only a one-way keyed hash of each sign-in identity (GitHub or Google and its user id, or your normalized email address) stays, with whether the free trial interview was used, so a new account with that identity doesn't get the trial again, and the keyed hashes the free sign-up credits' checks keep; nothing else about the account is kept. The database's point-in-time recovery history (Cloudflare D1 Time Travel) still holds them for up to 30 days, after which they are gone.
- Sign out everywhere ends every session, on every device, signed-in apps and the command line tool included (signing out of the command line tool ends just its own).
The server is the open source backend, so all of this can be checked in the code. Questions go to GitHub issues.