klog is a plain-text file format and a command line tool for time tracking. It is free and open-source software distributed under the MIT License.
Get the latest version of klog here: github.com/jotaen/klog
The idea behind klog is to store data in plain-text files in a simple and human-readable format. The notation is similar to how you would write down the information into a physical notebook using pen and paper. Manipulating your data is as easy as opening the file in a text editor and making changes to the text. By using the klog command line tool you can search, evaluate and manipulate your data from the terminal.
Let’s say you started a new job and want to use klog for tracking work times:
First day at my new job
8:30 - 17:00
-45m Lunch break
At the 24th of March 2018 you came to the office at 8:30 in the morning and went home at 17:00 in the afternoon. Somewhere in between there was a 45-minute lunch break.
As you see, klog supports different notations to record times. You can capture short summaries about your activities along with the time data (only if you want, of course), which can help you later on to make sense of what you did back in the day. And by adding tags you are able to run more fine-granular evaluations.
When stored in a file (e.g.
worktimes.klg) you can use the klog command line tool to
interact with it. For instance, you could evaluate the resulting total time like this:
klog total worktimes.klg
(In 1 record)
In case you want to add an extra 30 minutes, you can do:
klog track '30m' worktimes.klg
klog is based on plain text files that hold the data.
The file extension is
.klg file can contain any number of records
that each consists of a date, time entries,
and (optionally) summary texts.
First day at my new job
8:30 - 17:00
-45m Lunch break
8h15m Onboarding and introduction
Started to work on a first small
project (new sales brochure)
8:50 - 12:45
14:00 - 18:00
Records are separated by one blank line between them.
The first line of a record must be a date (formatted either
Entries are the actual time values that you track. Each entry represents an amount of time that you spent on something. The computed total time of a record is the sum of all its entries.
8:00 - 9:00
Both entries in this example are worth 1 hour each, resulting in a total time of 2 hours for this day.
Entries appear one per line and are indented by one level. Indentation is either 1 tab or 2–4 spaces. The order of entries can be arbitrary and has no influence on the result. Entries can be:
A duration, e.g.
12h48m. This can also be a negative value, in which case it will be deducted from the total time.
A time range, e.g.
12:32 - 17:20or
8:45am - 1:30pm.
The purpose of summaries is to capture arbitrary information alongside the data for future reference.
Woke up this morning with the plan
to make the world a better place
10h Watched movies
Summaries are optional and can appear:
- Underneath the date, in which case they are supposed to refer to the entire record. Record summaries can have multiple lines of text.
- Behind entries on the same line, in which case it is only supposed to refer to that very entry. Entry summaries are separated by whitespace from the preceding time value.
Tagging / categorising
Summaries can contain
#hashtags that allow for more fine-granular
filtering of the data later on.
Did some #sports
2h #badminton session with Max
1h Went out for a #run
If a tag appears in the overall record summary, then all entries will match when filtering for that tag. If a tag appears in an entry summary, only the respective entry will match.
In this example, the total time for the tag
#sports would be
and the total time for
#run would be
Open-ended time ranges
In case you just begin an activity (without knowing when it will end) you can already log it in your file as an open-ended time range.
Started to read my new book
16:30 - ?
Open-ended time ranges are denoted by replacing the end time with a question mark, otherwise they work the same as normal entries. Note that there can only be one open-ended range per record, and it doesn’t count towards the total time as long as it’s open.
There are use-cases where you have a certain overall time goal that you want to achieve. This so-called should-total property can appear after the record’s date, surrounded by parentheses. It is a duration value followed by an exclamation mark. For example, let’s say you are supposed to work 7½ hours per day:
8:00 - 16:00 Work
-45m Lunch break
Should-totals are a meta-property that can be useful for evaluation purposes, e.g. when you want to diff actual times against your designated goal. The should-total always applies to the entire record with all its entries.
Sometimes you start an activity in the evening and end it after
midnight, just so that start and end time don’t belong to the
same calendar date. For this case it is possible to “shift over”
a time to the previous or to the next day by adding the
< prefix, or the
> suffix respectively.
<23:30 - 8:00 Worked a night shift
22:30 - 1:45> Watched some movies
When filtering records, keep in mind that these entries are still
associated with the date they are recorded under, so the total time
for the above date
(If there are records for the adjacent days, their total time won’t be affected.)
Is it possible to use to-the-second precision, like
No, this is not supported. The reason is that it would effectively prohibit mixing values with and without seconds, which leads to a lot of hassle. Keep in mind, klog is for tracking time of activities, it’s not a stopwatch.
Can I capture timezone information?
No. In case you are affected by a timezone change or a switch to daylight saving time you need to account for that yourself. Realistically, this doesn’t happen all too often anyway, so it’s a tradeoff for simplicity to omit the timezone information altogether.
Can there be multiple records for the same date in one file?
Yes, as many as you want.
Are there any constraints in regards to the way I track times?
klog generally doesn’t impose too many restrictions.
Examples of things that are allowed:
tracking more than 24h per day;
tracking only negative durations;
time ranges that overlap;
negative should-total values;
empty records (that don’t contain any entries).
There might be valid reasons for such use cases, so klog deliberately is not stricter than necessary.
What the command line tool does do, however, is to raise warnings about certain potential issues, for example when it detects an open time range at a past date.
SpecificationYou find the formal specification of the file format in the repository.
Command line tool
The command line tool allows you to find and filter records in files,
pretty print and evaluate them, and apply some basic manipulations.
The following sections demonstrate the usage of the subcommands.
(They are not a complete reference, though.)
In order to learn about all flags and options please run klog with the
(which is also available on the subcommands).
In case there are syntax errors in the input text, klog cannot process it and displays an error message instead, which hopefully helps you fix the problem. Also some commands emit warnings to draw your attention to potential logical issues in your data.
For the evaluation commands you can specify input in one of the following ways:
- Pass one or more file names
- Use the bookmark functionality
- Pipe the data to stdin
Pro-tip: most shells have native support for glob patterns, so in case
you want to organise your records throughout multiple files
(e.g. one file per month)
you can process them all at once by passing the glob pattern
klog print worktimes.klg
total subcommand evaluates the total time
of the records from one or more files.
klog total sport.klg
If you want to evaluate all records in
since January 2018 that are tagged with
#workout, you would do:
klog total --since=2018-01-01 --tag=workout sport.klg
Note that open time ranges are not taken into account by default.
You can factor them in by passing the
klog total --now sport.klg
Prints a report of the records as a timeline (like in a calendar), listing all days and their total times.
klog report 2020.klg
Use with the
--fill option to print
a consecutive stream of all dates.
Displays an overview of all available tags and the aggregated total time for each of them.
klog tags sports.klg
Displays a side-by-side view of the most recent record (either today
or yesterday) and the overall statistics, similar to
klog total --now.
klog now sports.klg
In case you have
should-totals set this will also calculate the E.T.A. (the “estimated
time of arrival”, i.e. at which time your goal will be reached).
By using the
--follow flag you can keep the shell process open
to display live updates.
Add a new entry to an existing record:
klog track '1h30m' sports.klg
If you don’t specify a date then klog will look for a record at today’s date. Say, today was the 14th of February 2012, then the above example command would append a new entry so that the resulting record would look like this:
8:15 - 9:45 Morning run
You can not just track time values that way, but you can also add the summary text right away. (Effectively, the argument is just the entire entry line as it is supposed to appear in the record.)
klog track '8:00-16:00 Workday' work.klg
Note that you should wrap the value into single quotes in order for your shell to treat it as single argument and to avoid special characters to be interpreted. In case you want to track a negative duration please remember to escape the leading minus with two backslashes:
klog track '\\-30m Coffee break' work.klg
Starts a new open-ended time range:
klog start work.klg
The command expects a record to be present at the current
date. You can also specify a date explicitly through the
By default the current time is used as start time.
You can optionally specify it with the
--time 5:00 flag.
If you want, you can use the alias
Stops an ongoing open-ended time range:
klog stop work.klg
Otherwise it works in the same way as
klog start does.
Creates a new record:
klog create reading.klg
By default the current day is used as date.
You can also specify a date explicitly via the
The command assumes the records in the file to be in
chronological order and inserts it at the right position.
You can pass a should-total property with the
Configure a default file for klog to read from:
klog bookmark set my-current-file.klg
When a bookmark is set, you don’t have to specify an input file for the subcommands anymore. (You are still able to, though.) Bookmarking is a useful convenience so that you don’t have to browse to a specific working directory all the time. Instead, you can interact with your bookmark from any location:
klog total --today
You can also quickly open up the bookmark in your favourite editor
(based on the
$EDITOR environment variable).
klog bookmark edit
Converts records into a JSON data structure that allows you to process the data by means of other tools or scripts programmatically.
klog json somefile.klg
The output structure is not explicitly documented yet.
Run it with the
flag in order to explore the output structure, it should be
Remember that klog checks stdin for input as well, which might be handy when you invoke it as subprocess from another program.
widget (MacOS only)
On MacOS there is an experimental menu bar (systray) widget bundled into the command line tool. It displays an ongoing counter of the current day and the statistics of a file. As source file the bookmark is being used. You can launch the widget by running:
--detach option to run the widget standalone
(detached from the shell process).
Displays which version of the command line tool you are on. This also checks online whether there is a new version available.
This is an ongoing collection of all kinds of extensions around klog.
If you have built a utility for klog, please
open an issue
so that it can be added to this list.
In case you need the klog logo for anything, see here.
Editor plugin for VSCode for syntax-highlighting and snippets.
(Maintained by @vladdeSV)