A Brief History of Bitcoin DIY Hardware Signers: From Fork to Cross-Border Cooperation
I'm posting this for the author as he has not X Premium. It was written in chinese at first and then translated into English:
Author: @arh9992
YouTube: https://www.youtube.com/@ahr9992
The Beginning: If Trezor's Code Was Already Open Source, Why Not Build One Yourself?
The starting point of this history is a question that seems almost obvious, yet one that very few people ever actually acted on.
In February 2018, a developer going by "pitrezor" posted a tutorial on their blog with a title blunt enough to feel like a challenge to the entire hardware wallet industry: since @Trezor One's firmware was already fully open source, why couldn't an ordinary person build a functionally equivalent Bitcoin signing device themselves, using nothing more than a Raspberry Pi worth a few dozen dollars?
That was the starting point of the PiTrezor project. Its approach was distinctly "engineer-minded": rather than designing a new signing scheme from scratch, it took the official open-source Trezor One code as-is, wrapped it in a thin adaptation layer, and ported the code — originally written to run on Trezor's dedicated chip — onto the Linux environment of a Raspberry Pi Zero (or Pi 4). The hardware list was minimal: a Raspberry Pi, a 0.96-inch OLED display (or simply an HDMI output), plus one or two buttons. The project claimed 100% compatibility with the official Trezor web wallet, meaning it could be used to sign real transactions directly. It also took advantage of the Raspberry Pi's built-in hardware random number generator to strengthen key generation security, and optimized boot time down to around 5 seconds.
PiTrezor never pretended to be a "plug-and-play" product. The third-party review outlet WalletScrutiny explicitly classified it as a "DIY Project" in its analysis, with a specific caveat: the project distributed pre-compiled firmware, meaning users needed to independently verify whether that pre-compiled binary could actually be reproduced in full from the published source code. This turned out to be a shared challenge facing almost every DIY signing device project that came after it: open-source code by itself is not the finish line — "reproducible builds" are the last mile that actually makes good on the promise of open source.
BitBoy and Bowser Wallet: Teaching Toys Become Reality
If PiTrezor proved that "forking someone else's open-source design" was viable, the next two projects proved that "designing a DIY signing device from scratch" was equally viable — and, notably, their original motivation wasn't security anxiety at all, but a teaching experiment.
In September 2019, Justin Moon, an instructor at the Austin-based Bitcoin bootcamp BUIDL Bootcamp, released BitBoy: a self-built hardware wallet based on the M5Stack (an ESP32 development board), featuring statelessness and QR-code air-gapping, which briefly went viral on Crypto Twitter. Moon himself admitted it was simply a fun teaching project he put together less than a year after entering the space.
Close behind it came Bowser Wallet, released in 2020 by a developer going by the pseudonym "arcbtc" (Ben Arc). Ben Arc is better known for LNbits (an open-source Lightning wallet/backend) and his early contributions to the Nostr protocol. His core philosophy behind building DIY hardware was to encourage ordinary people to build things themselves, so they could gain more autonomy in how they use their Bitcoin devices.
Specter-DIY and DIY Jade: One "Regular Army," One "Side Project"
Around the same time as BitBoy, a more systematic and, ultimately, far more influential project began to take shape — one that was never just another weekend hackathon experiment, but the project that truly kicked off the entire DIY signing device movement. In 2019, the CryptoAdvance team, led by physicist-turned-developer @StepanSnigirev, released Specter-DIY, built on the STM32 F469I-DISCO development board. From day one, the project deliberately committed to being fully open source end to end. Specter-DIY ran on a high-end board with a touchscreen roughly on par with an iPhone 4, and it was the first to establish two things that would later become near-standard features across DIY signing devices: first, true full air-gapping via QR codes (the device has no USB or Bluetooth/WiFi physical connection to any networked device whatsoever); second, a "stateless" mode of operation — the private key is only ever loaded temporarily when needed, and is permanently wiped immediately after the operation completes, reducing the device itself to a pure tool fully decoupled from however you actually choose to store your key. Its companion desktop wallet software, Specter Desktop, was the first coordinator software to support pure QR-code communication, letting users freely mix and match DIY and commercial devices to build multisig setups.
From the very first day of the project, Stepan pulled all the underlying Bitcoin computation logic out into a separate, reusable code library called embit. At the time, this looked like a routine engineering decision — but six years later it would become the "common ancestor" of the entire DIY ecosystem: both @SeedSigner and Krux were later built on top of that same embit library. Even though the three projects differ wildly in philosophy, UX, and target users, they've always been seen as "siblings" because of this shared core codebase.
In January 2021, a more unexpected player entered the arena: the veteran Bitcoin infrastructure company Blockstream released @BlockstreamJade. While it can be purchased as a finished product, it carried DIY genes from day one — it can run on devices like the LILYGO T-Display, meaning you can fully order the parts yourself and assemble it yourself, sidestepping part of the supply-chain risk.
SeedSigner: From a $50 Prototype to the "UX Champion"
If Specter-DIY proved that DIY "could be hardcore," then the emergence of SeedSigner was the first proof that DIY could also be "genuinely pleasant to use."
The project was created by a developer who has remained pseudonymous. In December 2020, with the help of the embit library — and substantial assistance from Stepan himself — he built the first working prototype on a Raspberry Pi Zero board that cost just $5. The project quickly attracted new contributors and kept pushing out new features, many of which were later adopted industry-wide as de facto standards: the option to generate a new seed from camera-captured image data, the open SeedQR encoding standard, and the hand-transcribed QR code interaction pattern, among others.
The core hardware choice was also a clever piece of design: the project deliberately chose the version 1.3 Raspberry Pi Zero, which has no WiFi or Bluetooth module, to physically guarantee that the private key could never leave the device wirelessly. Paired with a camera and a display, all transaction data exchange happens entirely through QR codes scrolling on screen. The seed is only ever held temporarily in memory while the device is powered, and is wiped instantly on power-off.
Interestingly, the SeedSigner core team themselves have acknowledged that, of the three major DIY projects, they are technically the "least hardcore." But precisely because of that, the team has poured nearly all its energy into polishing a beginner-friendly interaction experience — which is exactly what made SeedSigner the one with the loudest Twitter presence and the most podcast exposure among the three. For a project with zero marketing budget, run entirely by volunteers, that kind of attention is itself a scarce resource.
Krux: Handing DIY Over to Off-the-Shelf Boards
SeedSigner proved the Raspberry Pi route was viable, but a Raspberry Pi still needs an extra display and camera module before it becomes a complete device. @selfcustodykrux took a different path from there: it leans directly on the already-mature ecosystem of Kendryte K210-series development boards, which come with the camera, display, and buttons all integrated into a single ready-made housing (such as the M5StickV, Maix Amigo, and others) — devices that require no assembly at all. You can buy one, plug it in, and flash Krux firmware straight onto it.
Krux was started by a single developer in 2021, but — much like Satoshi's own choice — he later handed the project entirely over to other contributors and stepped away completely. Today the project is led by odudex, and it's also built on the same embit library that Stepan originally created. In recent times, the Krux team has been the main force pushing for more advanced features to be added to embit.
ShieldSigner and Kern: New "Variants" the DIY Ecosystem Grew on Its Own
ShieldSigner comes from a developer known as @YTCryptoGuide (GitHub handle "3rdIteration"), and is currently the most successful fork of SeedSigner — he added Satochip smart-card support to SeedSigner, and even designed his own custom expansion board for reading the smart card. SeedSigner has long drawn criticism for lacking a secure element, while a Satochip smart card is itself a secure element that simply lacks a screen to interact with the user — combining the two struck many as a natural match. Interestingly, even in this space, the true pioneer was still @SpecterDIY, which had already offered optional smart-card support years earlier.
Kern (github.com/odudex/Kern) is an experimental new project built from scratch in C by Krux maintainer odudex. It's still in an early stage, but it's already generating considerable expectations — for instance, its QR-code scanning module may well be ported directly into the next generation of SeedSigner, one built on a microcontroller rather than a Raspberry Pi, becoming yet another example of code reuse across projects.
November 2025, São Paulo: Three Parallel Lines Finally Converge
If every step up to this point was the product of each project independently figuring things out on its own, then the most significant turning point on this timeline came in November 2025 in São Paulo, Brazil — where core members of the Specter DIY, SeedSigner, and Krux teams, alongside experts on Satochip and Jade DIY as well as self-custody educators, gathered together in person for the first time, meeting each other face to face and kicking off a level of collaboration far deeper than anything before.
The gathering was made possible by Lucas Ferreira, executive director of the Brazilian nonprofit Vinteum. He managed to convince every key figure to attend, including Stepan Snigirev himself — no small feat, given that Stepan had come close to burning out a few years earlier after an intense development sprint, and had largely stepped back from the Bitcoin development community since, to the point that some people assumed they'd never see him again. Vinteum handled all the event's logistics, funding came from the Human Rights Foundation, and the venue was Vinteum's own hacker house in São Paulo, Casa 21.
The most important item on the summit's agenda was the handover of embit, the core codebase shared by all three projects. Stepan's primary goal for the trip was to formally transfer maintainership of embit to a new group of co-maintainers — he didn't want to keep being the bottleneck for the entire ecosystem's progress, and understandably wanted to shed most of the maintenance burden. In the end, Specter DIY, SeedSigner, and Krux each nominated one representative, who together became co-maintainers of the embit repository — completing this symbolically significant "passing of the torch."
During the summit, the teams showed each other their respective technical work without reservation: Specter DIY presented the implementation details of its STM32 secure bootloader — a mechanism that ensures the device refuses to install any unsigned "evil firmware," arguably the single most important security guarantee a signing device can offer. SeedSigner's volunteer UX designer walked through his methodology for refining the interaction experience, which has already inspired Krux contributors to start experimenting with their own UX improvements. The Krux team, meanwhile, designed an encrypted multisig challenge that could only be cracked by making full use of its unique advanced features, demonstrating just how much complexity the tool can handle in the hands of an expert. The technical strengths each project had built up independently over years of working in isolation were, for the first time, laid out on the same table — and anyone is free to use any of it, no permission or license required, because it was all free and open-source software to begin with.
Closing: Why the DIY Spirit Still Matters
Looking back over this arc — from PiTrezor's first attempt at "forking open-source firmware" in 2018, to three teams formally handing over a shared codebase together in São Paulo in 2025 — a clear thread emerges: the old Bitcoin adage "not your keys, not your coins" has been pushed one step further by the DIY movement — it's not just the private key that you should control yourself, but the very device that generates and uses that key, which you should be able to understand, verify, and even build with your own hands.
From PiTrezor's almost self-evident question, to Justin Moon's teaching-driven conviction that "building a hardware wallet is the best way to understand private keys," to Specter-DIY, SeedSigner, and Krux each growing independently for seven years before finally shaking hands and sharing a single codebase in São Paulo — this entire arc has been driven by the same cypherpunk instinct for self-reliance: refusing to place unconditional trust in any single supply chain, vendor, or chip, and answering that with the only real solution — making the knowledge and the tools completely open, so that anyone willing to put in the work can verify things for themselves, and so that different teams, entirely free of commercial competitive pressure, can freely learn from and build on each other. The history of DIY hardware signing devices was never just a list of device names — it's the answer the Bitcoin community has given, again and again, to a crisis of trust: through open-source code, a few dozen dollars' worth of electronic components, and ultimately, through trust in each other.
[1]：https://github.com/kdmukai/article-diy-signing-device-summit/blob/main/article.md