Our Threat Model
Our threat model, stated plainly, is that nobody can touch your Lomenett vault without physically holding your specific device, and that the one attack with a real chance of working requires stealing it, a silicon lab, and more time than the theft buys. This page names every attacker we designed against, what each one needs, and what each one gets, including the ones we cannot stop.
Most companies keep their threat model internal. We publish ours, because a security model that only works while it stays secret does not actually work.
The ledger
| Attacker | What they need | What they get |
|---|---|---|
| A hacker anywhere on the internet | A path to the device, which does not exist. There is no Wi-Fi, no Bluetooth, no app, no account, and no server. | Nothing, because there is no path to attack. |
| Malware on a computer you use | To already control that computer. | The passwords typed on that machine while it watches, but never the vault, never your PIN, and never the entries you did not use. |
| Whoever finds or steals your stick | Your PIN. | A handful of guesses, counted in the secure element's own hardware, before the key is destroyed forever. |
| A professional with your stick and a lab | Weeks to months of work, lab-grade silicon attack equipment, and rare expertise, described below. | A small but real chance at your vault. |
| NuLenke (us) | A backdoor, which we did not build. The open-sourced firmware and schematics will let you verify that claim. | Nothing. We cannot open your vault, which means nobody who pressures, hacks, or impersonates us can open it either. |
The attack that could actually work
We will not pretend the professional's row is empty, so here is what a successful attack on a Lomenett looks like.
The attacker steals your specific device. Software attacks end at that point, because dumping the flash yields ciphertext with no key, and the PIN allows only a handful of hardware-counted guesses. The attack therefore has to go through the silicon itself, which means decapping the secure element, imaging it under microscopes, and attempting fault injection or microprobing against a chip that was designed to resist exactly those techniques. That takes weeks to months of work, equipment most attackers will never have access to, and rare expertise. A failed attempt can destroy the chip along with the data inside it, and none of the work carries over to a second victim, because the next attack starts over with a new stolen device.
Could it succeed? We assume yes, eventually, and our CEO's line is the whole philosophy: "Everything is hackable. It's just a matter of time. All we can do is extend that time." The chance is small, but it is not zero, and anyone who tells you zero is lying. What the design guarantees is that the only viable attack is the slowest and most expensive one, and that it only ever works against one person at a time.
Your half of the defense
The time that attack costs is your window. We built Lomenett to be "lose it and shrug", and we stand by that, but best practice is still best practice. If you are afraid someone else may have access to your device, change your passwords, starting with important accounts like your bank and your email, since email can reset most other accounts. Restore onto a replacement from your backup and rotate the important passwords in an afternoon, and at that point even the lab attack above is aimed at passwords you have already replaced.
The key destruction setting
The key destruction that stops PIN guessing is not a timeout or a lockout. The security silicon counts wrong attempts in its own hardware, and past the limit it destroys the key permanently, after which nobody, including us, can recover the vault. You can set how strict that trigger is, tightening the wrong-attempt limit if brute force is your main concern, or leaving more room if mistyping your own PIN is the more likely problem. Either way the counter lives in the chip rather than in software an attacker could patch.
Verified, not believed
We plan to open source all of it, meaning the firmware and the schematics, before production units ship. We encourage people to build their own from the files. We understand that not everyone has time for that, which is the reason we also sell them assembled. Either way, this threat model can be audited by anyone who cares to check, and we would rather be verified than believed.
The broader map of what is in and out of scope is at Our Attack Surface, and the question-by-question version is Can Lomenett Be Hacked?. The device all of this defends is at nulenke.com.