A Supercharged Keyboard Programming Daemon ⌨️
ktrl is a Linux keyboard programming daemon. It aims to aid you in the never-ending quest of achieving the ultimate keybinding setup.
You can dip your toes by remapping a few modifier keys (e.g
Ctrl). Or you can go all-in by creating a sophisticated layering setup with dual-function keys, tap-dancing, etc...
ktrl is heavily inspired by the amazing open-source keyboard firmware project QMK. You can think of ktrl as an attempt to re-implement QMK as a Linux daemon.
This is an alpha state project. If you find any bugs or quirks please reach out to me.
Table of Contents
ktrl sits right in the middle of the human-interface software stack. It lives in userspace, between the kernel and your display server (a.k.a X11 / Wayland).
This position allows ktrl complete control over the events your keyboard generates. These events are either transparently passed-on or transformed into ktrl's "Effects" (more on that later).
Aside from the obvious key remapping capability, here's a taste of some of the major things you can do with ktrl -
Although "layers" might seem like a foreign idea, it's something you're already very familiar with. After all, you apply "layers" all the time by using modifier and function keys :)
QMK takes this mechanism and generalizes it. Letting you design your own custom keyboard's layers!
If that sounds confusing, I encourage you to head over to QMK's documentation about layers.
Tap-Hold keys let you do one thing when the key is pressed, and another thing when it is held. For example, you can make your Spacebar act as normal when tapping, but serve as Ctrl when held.
Tap-dancing is quite similar to Tap-Hold. The key will act in one way if tapped once, and will act differently if tapped multiple times.
Again, both of these were shamelessly taken from QMK.
Hyperare special modifiers for creating keybindings that'll probably never conflict with existing ones. That's possible since
Hyperis equal to pressing
Mehis the same as pressing
Ever wanted to bind your favorite 8bit tunes to your key-presses? Well, now you can! Though, aside from making your hacking session more musical, this feature as some very practical uses as well.
For example, it can help you build new muscle-memory connections using audible feedback. See the Capslock <-> Ctrl example below for more on that.
Start off by grabbing the main
ktrlexecutable. Here's how you do that -
sudo cargo install --root /usr/local ktrl
Note: you may need to install
libtool. For Debian/Ubuntu distributions this can be done with
# apt install libalsa-ocaml-dev autoconf libtool libtool-bin
Although a bit cumbersome, this step makes sure we can run ktrl without root privileges. Instead of running it as root, we'll make a new user for ktrl. Then, we'll add the new user to the input and audio groups. Let's get started -
sudo useradd -r -s /bin/false ktrl sudo groupadd uinput sudo usermod -aG input ktrl sudo usermod -aG uinput ktrl
If you're using the sound effects
sudo usermod -aG audio ktrl
Now, let's add a new udev rule that'll allow ktrl to write to
/dev/uinputis ktrl's output device. Your keyboard being the input device.
git clone https://github.com/itaygarin/ktrl cd ktrl sudo cp ./etc/99-uinput.rules /etc/udev/rules.d/
Note that'll need to reboot your machine for the changes to take effect...
Now, it's time to decide where you'd like ktrl's assets and config to live.
By default, ktrl will assume you've placed these under
/opt/ktrl/assets. Though, you can override these defaults with the
To set-up the defaults, you can follow these steps -
# Asumming you've already cloned and cd`d into the ktrl project
sudo mkdir /opt/ktrl sudo cp -r ./assets /opt/ktrl sudo cp examples/cfg.ron /opt/ktrl
sudo chown -R ktrl:$USER /opt/ktrl sudo chmod -R 0770 /opt/ktrl
For ktrl to work, you have to supply it with a path to your keyboard's input device. Input devices reside in the
Linux provides two convenient symlinks-filled directories to make the search process easier. These directories are
Within these two directories keyboard devices usually have a
-kbdsuffix. For example, in my laptop, the path is
If you are using a bluetooth keyboard, you will need to locate the associated
/dev/input/by-*only lists usb connected inputs. The best way to do this is to run
cat /proc/bus/input/devicesand search for the associated bluetooth keyboard (the
Physoutput will have the MAC address). Here is an example:
I: Bus=0005 Vendor=05ac Product=0239 Version=0050 N: Name="My Keyboard" P: Phys=SO:ME:MA:CA:DD:RS S: Sysfs=/devices/pci0000:00/0000:00:01.2/0000:02:00.0/0000:03:08.0/0000:07:00.3/usb3/3-5/3-5:1.0/bluetooth/hci0/hci0:256/0005:05AC:0239.0009/input/input40 U: Uniq=SO:ME:MA:CA:DD:RS H: Handlers=sysrq kbd event25 ledsNotice in
Handlers, it tells us the event # we need. So our path to our device is
ktrl is a daemon that's designed to run as a background process. Therefore, you might want to set it up as a service. Feel free to skip this step if you just want to experiment with it.
Creating a service will vary from distro to distro, though, here are some basic steps that'll get you started on
systemdbased systems -
# Again, asumming you've cloned and cd`d into the ktrl project
edit ./etc/ktrl.service # change your device path sudo cp ./etc/ktrl.service /etc/systemd/system sudo systemctl daemon-reload sudo systemctl start ktrl.service
Finally, we get to the cool part! Though, let's briefly go over ktrl's config primitives before assembling our first config file
Specifically, ktrl uses a subset of the event codes that describe keyboard keys. E.g
KEY_LEFTCTRLdescribe the 'A' and Left-Ctrl keys.
Within a layer, we map a source key (ex:
KEY_A) into an
Action. Actions describe the physical input movements you'll apply to the source key. E.g A
TapHolddescribes a tap and a hold.
Tap: This is the default keyboard action. Use for simple key remappings.
TapHold: Attach different
Effects for a tap and hold.
TapDance: Attach different
Effects for a tap and multiple taps.
Actionwill contain one or more
Effects. These are the virtual output effects that'll manifest following the action. E.g Playing a sound, toggling a layer, sending a key sequence, etc...
NoOp: As the name suggests. This won't do anything.
Key: This is the default effect you're "used to".
KeySticky: Once pressed, the key will remain active until pressed again (like Capslock).
KeySeq: Outputs multiple keys at once. E.g
Meh: A shorthand for
KeySeq(KEY_LEFTCTRL, KEY_LEFTALT, KEY_LEFTSHIFT)
Hyper: A shorthand for
KeySeq(KEY_LEFTCTRL, KEY_LEFTALT, KEY_LEFTSHIFT, KEY_LEFTMETA)
ActivateProfile: Activates a user-defined profile
DeactivateProfile: Deactivates a user-defined profile
DeactivateAllProfiles: Deactivates all user-defined profiles
ToggleLayerAlias: When pressed, either turns on or off a named layer.
ToggleLayer: When pressed, either turns on or off a layer.
MomentaryLayer: While pressed, the relevant layer will remain active
Sound: Plays one of the pre-built sounds
SoundEx: Plays a custom sound provided by you.
Multi: Lets you combine all the above effects. E.g
ktrl uses the wonderful ron (Rust Object Notation) to make serializing configuration much easier. The format should be pretty intuitive, though please refer to the supplied cfg.ron for a practical example.
This is probably one of the most effective yet simple changes you can make right now. You're left pinky will greatly appreciate this change in the long-run.
Doing this with ktrl is easy. In one of your layers, add the following -
KEY_CAPSLOCK: Tap(Key(KEY_LEFTCTRL)), KEY_LEFTCTRL: Tap(Key(KEY_LEFTCTRL))),
Though, let's make this more interesting, shall we?
To make the transition smoother, let's add an error sound effect to the left Ctrl. This'll remind you you're doing something wrong -
KEY_CAPSLOCK: Tap(Key(KEY_LEFTCTRL)), KEY_LEFTCTRL: Tap(Multi([Key(KEY_LEFTCTRL), Sound(Error)])),
Ah, much better!
Of course, you can also go cold turkey and only leave the sound effect. Like so -
KEY_CAPSLOCK: Tap(Key(KEY_LEFTCTRL)), KEY_LEFTCTRL: Tap(Sound(Error)),
Another change I've been experimenting with is mapping modifiers to home row keys Though, due note you'll have to calibrate the
tap_hold_wait_timeconfig value to avoid false-positives.
Here's an example setup -
KEY_A: TapHold(Key(KEY_A), Key(KEY_LEFTCTRL)), KEY_S: TapHold(Key(KEY_S), Key(KEY_LEFTSHIFT)), KEY_D: TapHold(Key(KEY_D), Key(KEY_LEFTALT)),
This will make
Dact as usual on taps and as modifiers when held.
TapDancerequire calibration and tinkering. as stated above, you'll have to tweak the wait times for both of these to minimize false-positives.