Private by design.
Every connection is encrypted, every query is automatically scoped to your account, and every credential stays server-side — never on your device.
This page describes what's actually live in minoly today. Anything still in progress stays off it until it ships.
Protection already in production.
TLS encrypts all data moving between your device and minoly's infrastructure.
ProductionEvery database table enforces row-level security — your data is only ever visible to you.
ProductionPasswords are never stored in plain text. Google Sign-In never exposes your Google password to minoly.
ProductionAdaptive rate limiting and bot-detection reduce automated abuse across the platform.
ProductionAPI keys never touch the client app — they exist only in server-side infrastructure.
ProductionHow a request is protected.
Your Device
Every action starts here — opening today's tasks, saving a new goal, sending a message to Raa. Before anything leaves your device, the request is already being prepared for an encrypted connection.
Encrypted Transport
The request travels to minoly's servers inside a TLS-encrypted tunnel. Anyone watching the network in between sees only scrambled data, never the actual content of the request.
Backend Verification
minoly's server checks who's making the request and whether it's allowed to proceed — validating the session and running it through adaptive rate limiting before anything touches the database.
Database Row-Level Filter
The request reaches the database already scoped to your account. Row-level security means the query can physically only return rows that belong to you, regardless of what the application code asked for.
Response
Only your data comes back, over the same encrypted connection it arrived on. Nothing about the request or response is ever readable outside that tunnel.
Choose what others can see.
How adaptive rate limiting works.
Normal use
If you're a real person, you'll never feel this system exists.
The overwhelming majority of requests minoly sees look exactly like a real person using the app — opening today's view, checking off a task, sending a message. These pass through immediately, with no delay and no extra verification. Adaptive rate limiting is built to stay invisible during normal use; you should never notice it's there.
Fast bursts
A short pause, not a punishment.
When an account sends requests noticeably faster than a person typically would, the system introduces a short delay before processing them. This is enough to absorb an accidental burst — a double-click, a retried request — without punishing the account, while still slowing down anything moving faster than normal use ever does.
Suspicious patterns
Machines are consistent. People aren't.
Beyond raw speed, the system looks at the shape of the requests themselves — how evenly spaced they are, how repetitive they are, whether they look like they're coming from a script rather than a person. Genuine human use is naturally irregular; automated behavior tends to be unusually consistent. When a pattern looks more machine-like than human, minoly asks for extra verification before continuing, rather than assuming the worst outright.
Repeated abuse
A timeout, not a life sentence.
If an account keeps generating patterns that look automated or abusive after being slowed down and challenged, minoly applies a temporary block rather than a permanent one. The goal isn't to punish a genuine user who happened to trip a threshold — it's to stop sustained abuse while giving a real person a clear, fair path back in once the unusual activity stops.
AI works within clear boundaries.
Raa is built to be useful without being invasive. It's designed to help with what you're actually working on — not to quietly read through your entire account, not to remember more than a conversation needs, and not to treat your data as something to be used beyond generating your own responses. minoly's philosophy here is simple: an AI companion should only see what it needs to answer you, and you should always know, in plain terms, what that means in practice. Here's exactly what that looks like.
- Understands your goals, tasks, and notes. Raa can reference what you've already told minoly, so you don't have to re-explain your situation every time.
- Only pulls in context relevant to your specific request. It doesn't scan your whole account by default just because a conversation started.
- Never sends unrelated account information to the AI provider. What gets sent is scoped to what your message actually needs.
- Doesn't store your Google password or payment details. Raa is a conversation feature, not an authentication or billing system, and never touches either.
- You decide what you type. Treat sensitive details — passwords, financial account numbers, authentication codes — the same way you would in any chat: keep them out.
What happens when you delete your account.
Nothing happens without your say.
Deleting your account starts with you, deliberately. minoly sends a confirmation to your account email before anything happens, so a deletion can't be triggered by a mistaken click, a compromised session, or anyone other than you acting with intent.
Not just the easy parts — everything.
Once confirmed, the deletion cascades through everything tied to your account: your tasks, goals, plans, routines, notes, focus-session history, and your conversations with Raa. Each of those lives in the database because it's linked to your account, so removing the account removes all of it — not just the parts that are easy to delete.
Deleted means deleted.
There's no separate backup sitting somewhere with your data still in it, and no retention window where a deleted account quietly lingers. Once deletion completes, that's the end of it — not archived, not hidden. Gone.
Designed to resist abuse.
Pattern-based detection — It doesn't need to have seen an attack before to catch one.
minoly's abuse defenses don't wait for a specific known attack signature before reacting — they watch behavior. Request timing, repetition, and consistency are evaluated continuously, which means unusual abuse patterns can be caught even if they don't match something minoly has seen before. A real person's usage is naturally uneven; automated abuse rarely is, and that gap is what the system is built to notice.
Adaptive rate limiting — Your first line of defense, running quietly.
The same rate-limiting system described earlier on this page is minoly's first line of defense against abuse. Fast, automated, or repetitive request patterns get slowed down and challenged before they can do meaningful damage, while ordinary use — even fast, enthusiastic ordinary use — passes through without ever noticing the system is there.
Temporary, fair blocks — Abuse gets stopped. Mistakes get forgiven.
When abuse patterns persist past a warning delay, minoly applies a temporary block rather than a permanent ban. That's deliberate: it stops sustained abuse immediately, without permanently locking out a genuine user whose account triggered the system by mistake. The block lifts once the unusual activity stops.
Human follow-up — A person still checks what the system can't judge alone.
Automated systems catch most abuse on their own, but they're not the last line of defense. Reports and unusual account activity can be escalated for a real look by minoly's team, so patterns that are technically within the rules but still clearly abusive don't just slip through because a script didn't trigger.
Frequently asked.
Yes — every connection between your device and minoly's servers uses TLS encryption. There's no exception.
No. Row-level security is enforced at the database level — a query can only ever return rows that belong to your account.
No. Your notes, goals, and tasks stay yours. Content only leaves that boundary when you bring it into a conversation with Raa.
Raa sees your conversation and, when relevant, the specific context your request needs — not your entire account.
It's removed completely — tasks, goals, notes, routines, focus history, all of it. No backup, no retention window.