What is arbtt?

arbtt is a cross-platform, completely automatic time tracker.

There are lots of time-tracking programs out there that allow you to collect statistics about how you spend your time, which activities are your biggest time-wasters, and so on. However, most of them require explicit action on your part: you have to manually enter what activity or project you're working on, and that has several disadvantages:

  • You need to stop what you're doing to insert the meta-information, and that breaks your concentration;
  • If you are lazy or get annoyed and don't keep updating it, the statistics will be useless
  • You won't be able to catch a little thing like quickly answering an e-mail or looking for the weather report.

arbtt, on the other hand, is a time tracker that gets out of the way. Its core component (arbtt-capture) silently captures data about what you are doing, completely autonomously. No interaction required, no distraction possible. This information is continuously stored in a log file. A separate tool (arbtt-stats) the allows you to investigate this data, at whatever time is convenient to you, by using simple text-based rules.

One big advantage of this approach is that you do not need to know in advance what queries you are interested in. Since the rules are applied in real-time when you are evaluating your data, and not when recording it, your raw data is always intact, and you can add more rules and forgotten special cases later.

Keep in mind that the log file might contain very sensitive private data. Make sure you protect it well.
You can get rid of all logs by deleting ~/.arbtt/capture.log

How does it work?

arbtt-capture is a desktop daemon that runs in the background and keeps a continuous log of your activity, by storing at regular intervals which windows are open, which one has the focus and how long it has been since your last action.

From this log, a wealth of statistics can be derived. Here's where the "rule-based" part of the name comes along: arbtt comes with a built-in command-line statistics generator (arbtt-stats) that will, based on very simple but powerful rules you can customize, sift through the raw data and reveal patterns and relevant information.

The rules are specified in a simple text-based format, on a file called "categorize.cfg". Here's an example of a simple categorize.cfg file:

--Convert program executable names to recognizable names
aliases (
  "Navigator"         -> "Firefox",
  "evince"            -> "PDF reader",
  "gedit"             -> "Text editor",
  "totem"             -> "Video player",

--Mark any samples captured after 5 minutes of inactivity with the "inactive" tag
$idle > 300                     ==> tag inactive,

--Tag each program with its executable name (filtered by the aliases above)
                                    tag Program:$current.program,

--Tag each program with its window title (filtered by the aliases above)
                                    tag Title:$current.title,

$time >=  8:00 && $time < 12:00 ==> tag time-of-day:morning,
$time >= 14:00 && $time < 18:00 ==> tag time-of-day:afternoon,

And here's the corresponding output of arbtt-stats -c "Program":

Statistics for category "Program"
      Program:Firefox |  8d06h20m00s |      49.01
Program:Google Chrome |  3d15h24m00s |      21.60
  Program:Text editor |  1d04h04m00s |       6.94
        Program:Skype |    13h55m00s |       3.44
     Program:Terminal |     9h22m00s |       2.31
   Program:PDF reader |     6h10m00s |       1.52
      Program:Desktop |     5h13m00s |       1.29
 Program:File browser |     4h11m00s |       1.03
 (53 entries omitted) |    22h59m00s |       5.68

Install arbtt

Binary installations

  • Debian/Ubuntu:

    arbtt is available both in Debian's and Ubuntu's repositories.

    Install arbtt:

    sudo apt-get install arbtt

    Set it up to start automatically on system startup:

    cp /usr/share/doc/arbtt/examples/arbtt-capture.desktop ~/.config/autostart/

    Start the daemon manually, if you want it to start capturing immediately, rather than at the next system restart:

    (arbtt-capture &)

    Create a minimal categorize.cfg file to allow arbtt-stats to be invoked without errors:

    echo "{\$idle > 60 ==> tag inactive}" > ~/.arbtt/categorize.cfg
  • NixOS:

    arbtt can be enabled in your NixOS configuration via services.arbtt.enable, see the NixOS module for more configuration settings.

  • Windows:
    Download the latest arbtt-setup.exe and follow the setup process as usual. This will set up the capturing daemon, but note that there is no graphical interface. Extracting stats from the data requires running arbtt-stats on the command line. Run arbtt-stats --help for a quick reference; for more detailed information, consult the manual.

Source installations

arbtt depends on several libraries whose development versions (name often suffixed -dev) must be available or the compilation may fail with errors such as X11- failed during the configure step; these libraries include:

  • X11, for the X11 Haskell binding (possibly named libx11-dev or xorg-dev in your package manager)
  • PCRE3, for pcre-light (possibly libpcre3-dev)
  • XSS, for arbtt (possibly libxss-dev)

Fetch the latest revision using

git clone https://github.com/nomeata/arbtt

and run the following command to build the arbtt binaries:

cabal install --installdir=bin --install-method=copy --overwrite-policy=always

You will now find the binaries in the bin subdirectory, which you can copy, for example, to /usr/local/bin.

If you use GNOME or KDE, you can copy the file "arbtt-capture.desktop" to ~/.config/autostart/. If you didn't do a system-wide installation, you'll probably need to put the full path to arbtt-capture in the Exec line of the .desktop file.

Mac OS X installation

arbtt can run on Mac OS X systems. To compile it, you need to install the pkgconfig and pcre source packages using Homebrew or MacPorts. For Homebrew, you can execute:

brew install pkgconfig
brew install pcre

or for MacPorts:

sudo port install pkgconfig
sudo port install pcre

Then, you can compile arbtt using the usual cabal command, as described above.

If you use MacPorts and have a linking error, it may be because of a conflict between the system libiconv and the MacPorts libiconv. Execute the following cabal command to resolve the conflict:

cabal configure --extra-lib-dir=/usr/lib

Get in touch

Mailing list

The main communication channel is the arbtt mailing list, arbtt@lists.nomeata.de, which is used for both users and developers of arbtt. If you have questions about arbtt, want to report bugs, or just share your experience with it, please join. Also any contributions in the form of ideas, code or documentation is highly appreciated. To subscribe to it, visit the mailinglist page.

Issue tracker

In addition to the mailing list, we use the issue tracker of the GitHub repository.

Developement repository

The source code is written in Haskell, hosted on GitHub.

History and credits

Arbtt was created in 2009 by Joachim Breitner. This webpage was designed by Waldir Pimenta in 2013. The background image is Revolutionstaschenuhr by Onnahfarg; CC-BY