A series where author show how to use taskwarrior to implement GTD. This series requires some knowledge of GTD.

Part 1: Intro

Taskwarrior is very efficient list manager.

Taskwarrior base command format?
Taskwarrior usage: task <filter> <command-name> <arguments>

What is a filter in Taskwarrior?
Task command by default run on all your tasks. So filter is used to select specific tasks. For example: task +next will show all tasks with +next tag.

Which command names you can use in Taskwarrior?
Command name can be next, add, done, list, etc.

What is an argument in Taskwarrior?
Argument is used to pass additional arguments to command. For add command it is task description, task add <description> will add new task with given description. It’s one of the most basic command.

How to specify task “group” when you add it?
When you add any new task you can also specify project (proj:xxx) and tags. We use project because we usually split big tasks into projects of tasks. Also, you can specify tags by adding +<tag> to task description. Example:
task add Read chapter 1 of GTD pro:read.gtd +read +@home

What task command do?
Running task itself is equivalent to task next command. It’s shows most important of your tasks, based on pre-configured filter (priority sorting).

How to change specific task/tasks?
Each task had own ID. You can use it to refer to specific task. For example to mark task as done, you can use task done <ID>. You can find required task by task <keyword> command. You can also specify multiple IDs, separated by comma or a contiguous range of IDs: task 1,3,5 done or task 1-5 done. NOTE: I have alias t=task to make it shorter, so I can run commands by t ....

Part 2: Collection

Collection is a capture process. Where you can save all your incoming tasks, this allows free your mind from remembering all this tasks. Items in inbox can be represented as tasks which needs to be processed.

Can you describe basic GTD collection requirements?
Usual GTD’s collection requirements:

The inbox is where you capture ideas and tasks as they occur to you.
The transaction cost of putting something in your inbox should be as low as possible.
If your inbox is not at zero, it has to be immediately visible.
You should need as few inboxes as possible (1-3).
You are able periodically review your inboxes (weekly/monthly).

I use default project, “Inbox” (instead aliasing in tag like in original article). So any new task will be added to “Inbox” project.

My task add command aliased to ==ta==.

Urgency

To add urgency to a inbox task, we modify its urgency coefficient. You can do this by putting the following line in your ~/.taskrc file:

urgency.user.tag.inbox.coefficient=15.0

But usually it’s can be useful with different tags (not inbox).

Inbox in terminal’s prompt

Inbox in terminal’s prompt, skipped. I think it’s generate a lot of clutter.

Report

Custom report can be configured through ~/.taskrc file.

For example.

report.in.columns           = id,description
report.in.description       = Inbox
report.in.filter            = status:pending limit:page (project:Inbox)
report.in.labels            = ID,Description

To see this report, just type ==ti (my alias to task in)==. You can use this ad mental habit to quickly check your inbox in plain list representation.

Part 3: Tickling

This chapter explaining how to use tickler file with Taskwarrior.

Adding wait: metadata to a task ensures that the task won’t be visible until the date that you set this metadata to. For example, when you enter the following command, the task doesn't show up until the following Sunday:
tat monday Remind mind, did I completed Syd-GTD note?
This task use special tat function, to easily add tickle items.

tat () {
    local deadline=$1; shift
    task add +tickle wait:$deadline $@
}

This way the task will only show up in your inbox the following Monday. Of course, you can also use full dates like 2023-09-24.

To review tickle tasks use this command: ==t +tickle all==

Thinking it over

Personally, I often encounter a situation where I need to think something over and decide what to do with it later. This really doesn’t have to end up on a task list because then there’s the possibility that I’ll see it before I’m ready to think about it. For precisely that situation, I made the think alias:

It is important that you don’t ever think something over twice. You don’t want to create a snooze button! A good practical tip is to make sure that every think item is a simple yes/no question.

I think it’s useful, so I have this alias too.

alias think='tat +1d'

Part 4: Processing

You usually process items following GTD processing workflow.

Processing requirements?

Processing your inbox should be very easy to start with.
Processing is about making decisions, not about starting or finishing tasks.

To start processing your inbox, type ==ti== (alias to task inbox). You should now see your inbox items if you’ve followed this series up to this point. For every task you see there, go over the following questionnaire.

Processing inbox algorithm? When you get into the habit of asking yourself all these questions/steps, there should come a point where you don’t need this detailed answer anymore.

Item is actionable?

It’s not: Decide whether to keep note of it in your reference system, whether you want think it over of just delete it. (Remember, when using think, you should enter a simple yes/no question.) Then check off the task.
It is: go on.

Is there more than one action required?

Yes: Start up a new project with first task (at least one have +next tag). ta +next pro:new.project "Set up a new git repository for the fancy new project" Add as many tasks as you can possibly think of for this project. Then check off the task.
No: Decide what action the +inbox task represents and just create it with +next tag. ta +next "Call Jamie to schedule a meeting."

How long the action is going to take?

It will take no longer than two minutes: Do it right now. Then check off the task t 1 done.
It will take longer than two minutes: Estimate how long it would take to complete the action and go on.

Are you going to delegate the task?

Yes: Modify the task to be a waiting task and annotate it with instructions on how to check whether the task is done yet. Then delegate the task. Let’s say the task has ID n. task n modify project:wait
task n annotate "Check: Call Dave and ask about the car tires." due:+1w wait:+1d
No: Go on.

Defer the task task n done
task pro:gtd add Call Jamie to schedule a meeting.

Make sure the action is really an action when you use +next tag. It must be something that you can start doing ==immediately== without thinking about “how?”, “where?” or “why?” thanks to the thought you put into it at this moment.

Contexts

How to group tasks by physical place, where are you doing tasks?
When you finally get around to doing, it will be very important for your mental health to keep in mind the context in which you are working. Luckily, Taskwarrior makes it very easy to allow you to use your context to your advantage. For now, all you need to remember is to always add at least one context tag to a task. Common context tags are +@office, +@phone, +@computer, @shop, … Check more contexts variants in GTD contexts.

Research

Often the very next task for a project is just a lot of ‘looking into something’. For that reason, here is the ==rnd, Research and development== alias:

    # Usage:
    # rnd +next read first parst of XXX, create new task and do it in
    # nearst time
    # rnd check xxx book, check some book someday
    # t +rnd, list rnd tasks
    alias rnd='ta +rnd +@computer'

Read and review

Reading something later, when you know very well what it is you want to read is called ‘read and review’. Ideally, you know exactly what you need to do, which drawer you need to open, which file to open, which link to point your browser to. Taskwarrior allows you to keep this info in annotation with ==t annotate==.

Usually read and review happens on the internet. Enter the ==rnr <url>== to quickly generate this type of task.

TODO: dynamic include? To grab web page title I use complex url2text script, which utilize Brave browser (to support dynamic rendered web-pages, and avoid some false-positive web-sites protection issues).

And here rnr function:

# Usage: rnr <url>
rnr (){
    local text=$(url2text text "$1")
    local descr="\"Read and review: $text\""
    local id=$(task add +next +rnr "$descr" | sed -n 's/Created task \(.*\)./\1/p')
    task $id|grep Description
}

Part 5: Doing

Context

The most important step in choosing what task to work on is to realize that there are tasks you don’t even need to consider. If you’re on an airplane, and you have some time to work, there’s absolutely no way that you’re going to be able to water your office plants. Ideally, tasks that lie out of your current context should not even present themselves in the range of opportunities.

You can merge contexts into big one, or use custom parameters, for example (need place into ~/.taskrc):

context.work=+@computer or +@phone or +@online
context.home=+@home -@online

How to use contexts?
To enable contexts filtering (show only context-specific tasks), use context command:

task context home # At home
task context work # At work

Brain power

Currently, I don’t use this feature, but it’s interesting.

Here’s another example. You’ve just done an exam and your brain feels like its toast. There’s no use in seeing a task like Write a review of Marc's paper on monoids in the category of endofunctors (wink). A task like Water the office plants would ideally present itself at the top of your task list.

Can I create custom attributes for tasks?
You can use Taskwarrior’s user defined attributes (UDA) to accomplish this.

Taskwarrior supports a set of standard attributes for a task, known as the core attributes. These include project, description, due and so on. There are more than 20 standard attributes. They are necessary to provide all the functionality of Taskwarrior. For example, the project attribute is used to provide feedback on completion of a project, the projects command itself, and project hierarchy filtering. The project attribute has a lot of functionality associated with it, and this is why project is a core attribute.

Start by defining a UDA for brainpower:

uda.brainpower.type=string
uda.brainpower.label=Brainpower
uda.brainpower.values=H,M,L
uda.brainpower.default=M  # recommended to set defualt, to not forget to add it

You can then add the brainpower consideration to your next report:

report.next.description     = Most urgent tasks
report.next.columns         = id, start.age, entry.age, urgency, depends, priority, project,tags, recur, scheduled.countdown, due.remaining, until.remaining, description, brainpower
report.next.filter          = status:pending and +next
report.next.labels          = ID,Active,Age,Urgency,Deps,P,Project,Tag,Recur,S,Due,Until,Description,Brainpower
report.next.sort            = brainpower-,priority-,urgency-

This will sort the tasks on your list in a decreasing order by brainpower first.

Estimate

How I can organize tasks by time estimate?
Estimate useful to soft limit time on given task. Similar to the way we define the brainpower UDA, you can also define a UDA for the time estimated to complete the task:

uda.estimate.type=numeric
uda.estimate.label=Estimate

Using your UDA’s

How to create and list task with UDA (estimate)?
Same as standard attributes, but with own prefix. To create a new task or list them, you now specify a custom UDA:

    ta es:5 "Call Jamie and ask him why my office plants were dead after my holiday weekend."
    task next estimate:5

The next report should now show you a significantly smaller amount of tasks to choose from. Even better, you are sure that you only see tasks that you can actually make progress on within the next ten minutes.

Inom Turdikulov