Autokey: Desktop automation utility for Linux

Like many of you, I've been aware that there are several desktop automation utilities available for Linux, but until recently, I'd never used one. However, one of our readers sent me an email suggesting that I check out Autokey, so I did. (Thanks Keith) Essentially, Autokey lets you assign commonly used text to a hot key; then the hot key can be used as a shortcut for the original text. For example, I live in Albuquerque, and as you might imagine, this is tedious to type and is often misspelled. Wouldn't it be nice if all I had to do was press alt-a, instead? As a start, this is the type of thing that Autokey can do for us, and more.

Once we've defined a piece of text with Autokey, there are a couple of ways we can access that text. We can assign the text to a hot key, such as alt-a or ctl-alt-z. Alternatively, we can make the text available from a menu in the Autokeypanel at the bottom of the screen. Finally, we can define an abbreviation that can be used in place of the text. For example, a Perl programmer such as myself, might like to use the abbreviation, “sub” and have it defined as:

sub<tab>x {


This way, any time I started to define a new subroutine, Autokey would fill in a basic skeleton of a subroutine and move the cursor back up to the subroutine name.

As you can see, we're able to send special keys by using their names. Autokey has a long list of special keys it can send. We also have the ability to define abbreviations to be case insensitive or to only trigger on windows with a specific name.

If this was all that Autokey could do, it would still be a useful tool, but this article wouldn't make for very interesting reading. Fortunately, there's more; Autokey has a built-in Python interpreter and an API that allows us to do some really interesting things.

He're a quick example:

choices = ["konqueror", "firefox", "chrome"]

retCode, choice = dialog.list_menu(choices)

if retCode == 0:

system.exec_command(choice + " " + clipboard.get_selection())

This quick little script allows us to highlight a URL, press a hot-key, select one of three different browsers from a dialog box, and open the URL with the chosen browser.

So let's dig a bit deeper into Autokey's Python API.

Autokey's keyboard class provides send_keys(), and a few variations, which provides a means of sending keystrokes to the windows manager. An Autokey script first calculates what it wants to send, and uses the send_keys() method to send it. The trick, of course, is to determine what to send, and that is the purpose of much of the rest of the Autokey API.

The mouse class provides click_relative() and click_absolute(). These methods simulate mouse clicks and allow us to determine which mouse button to simulate, as well as where it should click based on XY coordinates. The absolute variant simulates mouse clicks anywhere on the screen. The relative version simulates clicks within the currently active window.

The store class allows our Autokey scripts to persistently store key/value pairs and access them later. Here we find set_value(), get_value(), and remove_value(). The set_value() method accepts a key and a value. Then, the get_value() method can be passed a key value and will return the appropriate value. The remove_value() method deletes a given key/value pair from the database. I found it strange that there seemed to be no way to know if a particular key/value pair exists. Then I came across the has_key() method in one of the example scripts; it isn't documented in the API documentation. But, it apparently returns a boolean value indicating that a given key exists.

The system class provides only two functions, exec_command() and create_file(). The exec_command() method allows an Autokey script to run an external command and optionally capture the command's output. The only caveat is that the method, and thus Autokey itself, blocks while the command runs, so the command needs to terminate immediately. The create_file() method creates a given file and inserts content into it. Strangely, there doesn't seem to be a corresponding read_file() method.

The QtClipboard and GtkClipboard classes provide access to the windows manager's clipboard. As you can see, there is a Qt and a Gtk variant. You'll have to be sure to use the appropriate version depending on which windows manager you use. Both versions export the same methods. Here, it's important to recognize the difference between a clipboard and a mere selection. With this distinction in mind, we find that the *clipboard classes export get_selection(), fill_selection(), get_clipboard(), and fill_clipboard() methods. With these methods, we can get and replace text in either the clipboard or the currently selected text.

The engine class provides methods that allow us to create new scripts and abbreviations. I actually didn't find these methods too interesting myself, so I won't spend much time here discussing them. I simply appreciated their existence from a completeness point of view.

The window class provides methods for managing and manipulating windows within the windows manager. These methods allow us to move, resize, and close windows as well as move them to different desktops. We can also minimize, window shade, and maximize windows with methods from this class. While recognizing the power to be had from these methods, I don't envision the need to manipulate a particular window and thus, haven't explored this class very thoroughly. Suffice it to say that there is quite a bit you can do, if you need to.

I happen to think that I've described some pretty powerful Autokey capabilities, but here comes the really cool part. The dialog class, which also comes in both Qt and Gtk variants provides a very powerful mechanism for interacting with the user. Here we find methods for presenting the user with dialog boxes, password boxes, menu dialogs, file and directory pickers, and color pickers. This is where I think the real power is to be had from Autokey as it gives us the ability to write simple interactive scripts and assign them to hot-keys for convenient access. It wouldn't be hard, for example, to write a script that would allow us to select a URL from our browser window, press a hotkey, and be presented with the user name and password that we need to use to access the site. This password management system would work no matter which browser we chose to use.

Over the years, I've grown accustomed to having my windows organized in a particular way, depending on what I'm doing. For example, when I'm doing web development, I like a web browser occupying two thirds of the width of my screen, and all of the height. Then I like my console filling the rest of the screen. With Autokey, this configuration, and much more complex configurations, could be a keystroke away.

As a final example, I'll confess that I really enjoy watching my favorite TV shows on, with one minor inconvenience. Anytime I want to pause a show, I have to use the mouse, and I really find the mouse distracting. I have hot-keys that allow me to change the volume from my keyboard. I can flip between active windows and active desktops using nothing but the keyboard. I use the vi editor because it doesn't require me to use a mouse. Autokey would allow me to create a hotkey that would simulate a mouse click in the center of my hulu window, thus pausing and resuming the video.

I know this makes me sound lazy, but that's the point. The whole purpose of tools like Autokey is to automate menial tasks. After all, that's what technology is for; it's why our TV's all have remote controls.

Load Disqus comments