- Get a $50 laptop, wipe it with a clean install of Linux and keep it air-gapped.
- Roll some actual dice, generate the check sum in terminal, then look up the BIP39 words for your seed.
- Install Sparrow Wallet via a USB stick and create a hot wallet (software wallet) with the seed.
- Add a strong passphrase (12x BIP39 words).
- Export a watch-only wallet into Blue Wallet or Electrum for receiving addresses.
- Import TX history back into air-gapped laptop for signing.
- Broadcast via QR code on any push_tx website.
- Keep seed and passphrase backed up in separate locations.
High level security26.3%
Medium level security47.4%
Low level security26.3%
19 votes \ poll ended
Security is a process. Not a set of tools. So take away all the tools at first, don't build around them. Make a great process. Then, find the tools that will (provably...) make the process more bearable and check the tradeoffs.
What you just did here is to avoid a hardware wallet, let's trust cheap and definitely bugged hardware and install the software with the most exploits that ever existed: the linux kernel. Have you done a kernel review lately? It's really complex and it's rather hard to actually spot weak points and patch or work around these. Not to mention you didn't explain how you're going to patch your installed kernel in the first place. Easier to maintain reviews of an ESP32 firmware, because it is less complex.
If the laptop never touches the internet what bugs should I be concerned about?
Most of them. A physical attack doesn't need your device to be online.
If you're a nobody and/or you're protecting a small amount of coin, fine. If one day your stack becomes significant (from your subjective pov) or you grow up to be on anyone's radar, you need to reassess every choice you made based on available tech rather than based on protocol. But if you can't afford a $5000 laptop and need to get a $50 one, then that says something about your stack size already, so no worries, I guess.
We often use to model our security around available/known tooling, so this is not criticism of you or your solution per se; hobby security is the norm anyway in this space. It's fine if nothing is truly at stake.
Could be easy to fuck it up. Make sure you look into the dice rolling properly and follow a methodology that actually generates the level of entropy you need.
If you roll dice 100 times I can't see what the difference between common dice to casino grade.
Not dice, I mean the process to generate a raw seed and checksum / turning that into the BIP39 words ... if you know what you are doing cool, but it is not so simple.
Have a look at #1543613
Your process is generally sound, take a look at DTails, "a tool used to remaster a Tails/Debian based OS to add curated privacy and Bitcoin tools". Can skip a step or two above.
I've used Tails before. What's the advantage of using DTails? Is it simply more focused on bitcoin?
Im not the author -- but the author is on here and posted that link the other day.
My take is it allows you to inject Bitcoin tools into a tails image in a reproducible way -- which means you have everything you need to do keygen on an airgapped laptop.
Need to run your own node, or what good is the watch only wallet?
You need to get TX info like the amount of sats on which address.
Running a node on a separate device though it's not really necessary.
I don’t like creating a hot wallet with the seed
I don’t like typing the seed phrase into anything but a hww
Why so reluctant for a hww? Why no multisig?
U really trust an old laptop more?
I'm reluctant to put too much trust into one hardware wallet.
You can use multisig with multiple hardware wallets but then you lose some privacy onchain.
You could also use a multisig taproot setup for better privacy, but then there's a risk with quantum computers approaching.
How do you know if you are buying loaded dice? Make sure to use old Yahtzee or Monopoly dice? :D
Too many things that can go wrong with this...
It's fine. Basically a vanilla paper wallet approach.
Air-gap + dice + checksum is legit, that's the part almost nobody actually does right. The broadcast step is the leak though: pushing the raw tx through a random push_tx website hands that site your IP tied directly to the address you just funded, before it even confirms. Sparrow can broadcast through your own node or over Tor - worth doing instead of a public form for anything you actually care about.
Solid process, honestly better than half the hww theater out there. But the passphrase is the part that'll bite you: 12 BIP39 words as a passphrase has no checksum, so one typo makes a completely different wallet with zero warning. Wipe and re-derive from your written copy before real sats touch it