Everyone mentions using 6-sided die but this method I found is much easier/quicker and has more than enough entropy: https://thesimplestbitcoinbook.net/wp-content/uploads/2025/03/FINAL_Roll-Your-Own-Seed-Workshop_8.5x11_03_25_2025.pdf
It includes printouts you can use to write down the dice rolls and a color coded BIP39 wordlist to make it very quick and easy to identify the correct word for the roll.
I checked the above link and it looks good. However before doing anything what would you recommend in my case. As a result of the coldcard tragedy I am looking into the security of my main wallet a Trezor 5 (bitcoin only wallet). Yes we use a seedphrase generated by Trezor and no we do not use a passphrase. We also have a empty dormant Jade. As a non tech it is hard to distinquish in the tsuanami of what is the best practise now?
The thing that keeps nagging me is the timing of this breach, just 3 days after the Kimi K3 model weights were released. The actual shape of the CC bug was quite subtle and particularly easy to overlook. Multiple devs have since pointed the latest Opus model at the code base, and it couldn’t find the bug without hand-holding and giving it hints. And of course Fable refuses to even try. It suggests a scenario where AI that could be leveraged to spot an obscure bug like this just needed to reach a particular level of competence.
@Dennis66 I bring this up in response to your question, because if my suspicion is correct, then this is a warning shot that we are likely entering a new reality where obscure bugs that nobody spotted for years in open source projects will suddenly start being found and exploited. My prediction (which I really hope I am wrong) is that CC will not be the last hardware signer to be hit.
So what can be done, since we can’t know in advance the specific nature of these yet unknown hidden bugs? I think it has to be approached by adding multiple layers of security, so that a breach which takes out one layer shifts the attacker’s burdon to also cracking another layer (buying you time to move your stack while the main attack is focused on softer targets). We saw this with the CC bug, where having a strong passphrase on top of the seed phrase bought people more time to respond.
So some things that can be incorporated (all come with trade-offs, so you have to taylor your specific layers of security to your own situation)
Use physical methods to generate entropy when producing your seed phrase. These could include physical dice rolls or coin tosses, or photos capturing 2 or more complex system states (like lava lamps, Newton’s cradles, double pendulums, etc)
Adding a strong passphrase. This should be more than six words if using real words. This one itself has layers that can be added to it, like the software generating the words collecting entropy from mouse movements, generating the words on an air-gapped system, etc. I’m planning to publish a couple tools to help with this soon.
Entering the seed phrase only when it is needed, and then wiping the seed from the wallet between uses.
Multi-vendor mulisig, which would require two vendors to be compromised simultaneously for you to be hit, and stacks well with any of the above practices.
Paul, thank you that make sense. Although I only understand 50% of what you say we will use a dice method and create a passph. with 8 words. I don’t get the multi vendor solution. For example Jade created a multisig wallet where they have 1 key to my wallet!? This sounds suspisous to me now. After de CC debacle and everything that comes out on NVK, I have lost all my trust in the wallet company’s. Ordered some parts on Ali to build a ‘seedsigner’ that seems to be a good option for the security of our sats.
I meant a 2 of 3 mulisig, where all three keys are owned by you (no third party), and you use three different hardware wallet vendors for each key. For example: a Trezor Safe, a SeedSigner, and a BitBox02.
I will look into it.