The project positions itself explicitly as a launcher-based platform rather than another single-app replacement. By keeping the stock reader intact and adding a home screen that hosts multiple installable apps side by side, Cobalt argues the shift is from hobbyist hacking to consumer-grade multi-app UX.
Argues the distinction that matters is not that code runs on Kobo — that's been possible for a decade via KOReader, Plato, and custom firmware — but that Cobalt is the first attempt at a true app platform on cheap e-ink hardware. Collapsing install ceremony into 'install once, then browse apps' is the UX shift that unlocks casual experimentation.
Contrasts Kobo's open sideload path with Amazon's locked-down Kindle bootloader, noting Rakuten never bothered locking things down — you just mount over USB and drop files in a .kobo folder. This permissive stance from the vendor is what allows a project like Cobalt to exist as a sideloaded package on stock firmware rather than requiring a jailbreak.
Frames the historical Kobo scene as fragmented: owners either stayed on stock firmware or committed fully to a KOReader-style replacement, losing native Kobo store integration in the process. Cobalt's contribution is providing a middle ground where the stock reader coexists with third-party apps, removing the commitment cost of trying something new.
Cobalt, a project from BandarLabs, hit 496 points on Hacker News with a simple pitch: your Kobo e-reader can run apps now — not one hacked binary at a time, but a real launcher with multiple installable applications side by side. The project ships as a package you sideload onto stock Kobo firmware, after which a home screen sits alongside the native reader UI and hosts whatever you install: RSS readers, note-taking tools, terminal emulators, offline Wikipedia, a browser, chess.
The distinction that matters is not "you can run code on a Kobo" — people have been doing that for a decade — but that Cobalt is the first attempt at a coherent app *platform* on e-ink hardware you can buy at Target. KOReader, Plato, and the various Kobo Custom Firmware forks all exist, and all of them are essentially single-app replacements: you swap the reader for a better reader. Cobalt keeps the stock reader and adds an app model on top.
The technical approach is unglamorous and that's the point. Kobo devices run a stripped Linux userland on top of a Freescale/NXP i.MX SoC. Rakuten never bothered locking the bootloader the way Amazon did — mount the device over USB, drop a `.kobo` folder in the right place, and the firmware runs your code. Cobalt uses that same sideload path to install a runtime that manages processes, handles the e-ink refresh quirks (partial vs full refresh, ghosting mitigation), and exposes a launcher.
Kobo hacking has historically been a hobbyist scene: individual devs porting individual tools, each with its own install instructions, each fighting the e-ink refresh model from scratch. The result was that Kobo owners either lived with the stock reader forever or committed to a full firmware replacement like KOReader and lost the native Kobo store integration. There was no middle ground and no way to casually try a second app.
Cobalt collapses that ceremony into "install once, then browse apps" — the exact UX shift that took smartphones from Palm-era curiosities to platforms. That's a bigger deal than it sounds. E-ink hardware has been stuck in a strange local minimum for years: the panels have gotten dramatically better (Carta 1300, faster refresh, color via Kaleido 3), the SoCs are more than capable of running modern software, and the battery life is measured in weeks. But the software model has been "reader, and maybe a browser Amazon lets you use." The bottleneck has been platform, not silicon.
Compare this to the Kindle side. Amazon has spent a decade progressively locking down the Kindle: signed firmware, encrypted partitions on newer models, the jailbreak community reduced to a cat-and-mouse game against each firmware update. There is no serious third-party app ecosystem on Kindle and there will not be one. Kobo, by contrast, has quietly stayed friendly to modification — partly because Rakuten's model is selling books, not hardware lock-in, and partly because they simply don't have the engineering budget to fight it.
The HN thread reflects this: the top comments aren't about Cobalt's technical novelty, they're about *finally*. One recurring note is that people bought Kobos specifically because they suspected this day would come, and Cobalt is what they were waiting for. A second recurring note is the reverse — people who already own a Kindle asking, effectively, whether the next e-reader they buy should be a Kobo purely on this basis. That's the economic shift: for the first time, "can I run software on it" is a real feature that pushes hardware buying decisions in e-reader-land, the way it does in every other category.
If you're a dev who's ever thought "I'd love a distraction-free e-ink device that runs *my* tool," the calculus just changed. A used Kobo Clara HD runs about $60. A new Kobo Libra Colour is $220. Both are Cobalt-supported. That is materially cheaper than a reMarkable ($400+ with the subscription upsell), a Boox ($400-800 for Android-on-e-ink), or a Daylight ($729). And unlike reMarkable and Boox, you're not shipping into a vendor's app store or dealing with Android's e-ink refresh handling — you're shipping a Linux binary into a runtime that already knows how to talk to the panel.
Practical implications, in order of immediacy: RSS/read-later apps are the obvious first port — this is what a lot of us actually want an e-ink device for, and the existing readers are all either paywalled or coupled to the vendor. Note-taking with sync is next; the stylus-equipped Kobos (Elipsa, Sage, Libra Colour) become interesting targets if you can pipe notes to your own backend instead of Rakuten's. Terminal/SSH clients are a novelty for most people but a real workflow tool for anyone who does on-call from a couch. Offline reference tools — dictionaries, docs, Wikipedia dumps, code snippet browsers — are a genuinely good fit for the hardware in a way they never were for tablets, because you're not fighting a backlit screen or a 6-hour battery.
The distribution story is what makes this different from every previous "just install Linux on it" pitch: there's a launcher, so users don't have to be developers to try your second, third, fourth app. That's the entire ballgame. Kindle jailbreaks failed as a platform because installing anything was a 20-step process per app. If Cobalt keeps the install friction near zero, it can support the kind of long-tail apps — three-hour weekend projects, single-purpose utilities — that make a platform interesting.
One caveat worth naming clearly: Cobalt is early. Sideloading is unofficial, firmware updates from Rakuten could break the runtime, and there's no signing/sandboxing story yet, so "apps" are effectively trusted binaries. Treat it as a hobbyist platform for now, not a production target. The Kobo forum threads about broken installs after Rakuten firmware pushes are a recurring theme, and Cobalt inherits that risk surface.
The interesting question isn't whether Cobalt succeeds — it's whether Rakuten notices and pushes back. Amazon's lockdown of the Kindle happened *because* the jailbreak scene got interesting enough to matter to the business. If Cobalt gets the traction the HN reception suggests it might, expect a Kobo firmware update within a few quarters that tightens the sideload path. The window where you can buy a $60 used Kobo, install Cobalt, and treat it as a general-purpose e-ink device may be narrower than it looks. Which is, incidentally, an excellent reason to ship whatever app you've been meaning to build for e-ink now, while the platform is still open.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.