← All posts
Blog

Rebind: one programmable input layer, from the clinic to the clutch

Shout-out: Rebind

On my desk there's a tiny board — a Teensy 4.0, a thumbnail-sized 600 MHz microcontroller — flashed with firmware from Rebind. My keyboard and mouse plug into it; it plugs into the PC. To Windows, it is the keyboard and mouse. But in between, every keypress and every mouse delta runs through code I control before the OS ever sees it. That's the whole idea, and once it clicks, you start seeing uses for it everywhere — from a locked-down clinic workstation to a fighting-game arcade stick.

What it actually is

Rebind is a programmable input layer that lives in hardware, not software. Most macro tools — AutoHotkey, vendor software like Razer Synapse, Steam Input — sit somewhere in the operating system, which means they need to be installed, they need permissions, and anything watching the machine can see them. Rebind sits one layer lower: it's a USB device that intercepts input at the firmware level and re-emits a clean, standard HID stream. The tagline says it plainly — "your input, intercepted at hardware level." The OS only sees the result.

Three properties fall out of that design, and they're the reason it's interesting:

The dashboard makes it approachable

The part that surprised me is how low the on-ramp is. You don't have to live in the scripting layer to get value — there's a web dashboard where you upload macros, bind keys, and manage profiles, and you're productive in minutes. The depth is there when you want it (you can drop into Luau, talk to the device over a remote protocol from TypeScript / Python / Rust, record and replay sequences), but day one is "drag a macro in and go." That combination — trivial to start, deep when you need it — is rare in hardware like this.

Why "looks like a normal keyboard" is the killer feature

It's easy to read the spec sheet and think the headline is the 8 kHz polling. It isn't. The headline is that it needs nothing from the host machine. That single fact is what unlocks the use cases most people never consider, because it lets a programmable input layer go places software automation legally and practically cannot.

From the clinic: input where you can't install anything

Healthcare IT is locked down for good reasons — regulated, audited, no unsanctioned software on the endpoints touching patient data. That's exactly the environment where a hardware input layer shines, because there's nothing to install. The device is a keyboard. Hospital IT policy has no opinion about a keyboard.

Concretely, where this earns its keep:

None of that touches the protected system's software stack. It rides in on the USB port like any other peripheral.

Accessibility: remapping the human-to-machine seam

The same "it's just a keyboard, but smart" property is a genuinely good accessibility primitive. Because the logic is yours, you can reshape input to fit the person instead of forcing the person to fit the input:

Software AT (assistive tech) is great until you hit a computer where you can't install it. Hardware AT travels.

And the fun part: gaming — including the exploits

This is where most people meet a device like this, and I'm not going to be coy about what it does. Because the OS — and therefore the anti-cheat — sees nothing but a standard keyboard and mouse, a hardware layer can pull off things that software simply can't, with nothing on the machine to inspect:

Here's my actual position, and it isn't the cautious one. I think this stuff should be poked at, out in the open. I'm a white-hat at heart: the value of an exploit isn't in hoarding it, it's in showing it. Every undetectable hardware trick demonstrated in daylight is pressure on the defensive side to get smarter — better behavioral detection, input-provenance checks, server-side sanity that doesn't just trust the client. The cheats and the anti-cheats are two halves of one long, genuinely fun cat-and-mouse, and that game only gets better when people are willing to play the attacker's side and publish what they find.

Hiding the capability doesn't make it stop existing — it just means the only people wielding it are the ones who'd never tell you. So I'd honestly rather more people explored exactly what a Teensy in the line can do, because the next generation of anti-cheat should be built by folks who've already stared the problem in the face. Break it, understand why it works, show your work. That's how the workarounds after this one come out smarter — and that loop, to me, is the whole sport.

The caveats

Bottom line

What I keep coming back to is that Rebind isn't really a "gaming gadget" or a "medical tool" — it's a primitive. It's a fast, OS-invisible, programmable seam between a human and a computer, and it asks nothing of the computer to work. Point that at a clinic terminal and it's a compliance-friendly productivity multiplier. Point it at a game and it's an exploit worth dragging into daylight so the defenders catch up. Point it at an accessibility need and it's freedom. Same little board on my desk, doing all three. Shout-out to the Rebind team for building the layer and then handing you the keys to program it.