3rd-Party Application Guidelines

Taskwarrior can be extended by means of a third-party application, which can be a wrapper, or hook script. There are script examples of import and export add-ons that support many different formats (clone the repository, look in task.git/scripts/add-ons). Then there are more sophisticated applications such as Vit that provide a complete replacement UI.

All of these provide interesting new features and improve ease of use for different kinds of users. We encourage you to create such add-ons, but in doing so, there are some rules that must be followed, which will not only protect the users data from mistreatment, but also your application from being sensitive to changes in Taskwarrior.


  • Produce, consume and handle UTF8 text properly. UTF8 is the only text encoding supported by Taskwarrior.
  • Don't attempt to parse the pending.data file. Here's why: the .data file format is currently on its fourth version. The very first version was never released, so if you want to read Taskwarrior data properly, you will need to parse the three supported formats. Those formats are not documented. Additionally, you will need to handle the GC operations, implement the task "unwait" feature, observe user defined attribute handling restrictions, and implement recurring task synthesis all of which require .taskrc and default value access. You would essentially be rewriting the data access and configuration portion of Taskwarrior, which is a major undertaking. To support filters you would also need to evaluate the supported clauses, provide DOM access and implement aliases. Then there is also the fifth data format, which is planned...
  • Use the export command to query data from Taskwarrior. The export command implements filters which you can use, or you can omit a filter, get all the data, and implement your own filtering. JSON parsing is very well supported in all relevant programming languages, which means you should be using Taskwarrior itself to query the data, with a commodity JSON parser in conjunction. While the JSON format will be tweaked over time, the general form will not.
  • Beginning with version 2.4.5, use the import command to import modified JSON back into Taskwarrior.
  • Older versions: Use the command line interface to put data into Taskwarrior. Composing a valid command line is a simple way to put data in to Taskwarrior.
  • Use the _get helper command to access individual data items. Note that multiple items can be retrieved by one command. If accessing more than just a few items, use the export command.
  • Verify feature support by running task --version. This command returns the version number, which will help you determine whether or not a particular feature is supported. Note that this command does not scan for a configuration file, and is therefore safe to run if Taskwarrior is not yet set up.
  • UDAs (User Defined Attributes) must be preserved in the data. When reading the JSON for a task, there may be attributes that you have never encountered before. If this is the case, you must not modify them in any way. This not only makes your application future-proof, but allows it to tolerate UDAs from other data sources. It also prevents the Taskserver from stripping out your data.


  • If you need to store additional data, consider putting your own data file in the ~/.task directory. Just don't use the file names pending.data, completed.data, backlog.data, or undo.data.
  • There are many helper commands designed to assist add-on scripts such as shell completion scripts. These commands all begin with an underscore, see them with this command: task help | grep ' _'.
  • Familiarize yourself with the means of forcing color on or off, disabling word wrapping, disabling bulk operation limitations, disabling confirmation, disabling gc, modifying verbosity and so on. There are ways around almost all the restrictions, and while these don't make sense for regular users, they can be critical for add-on authors.