Writing Scripts

Interaction with external tools from a Kakoune session is supported
through the use of scripts. Once loaded by the editor (either
automatically or manually), they allow extending the functionalities
provided by default through commands and hooks.

Their implementation should be kept as simple as possible, as they are
not meant to be generic tools themselves but a mere API to actual

                                          +--> | tmux |
                                          |    +------+
        +---------+                       |    +------+
        | Kakoune | <------->  scripts  +-+--> | X11  |
        +---------+                       |    +------+
                                          |    +------+
                                          +--> | ...  |


The amount of dependencies of a given script should be kept to a minimum
for practicality reasons, and have to be reasonable and expected
considering the purpose of the script itself.


  • the clang.kak script provides with code completion using the
    clang compiler

  • the tmux.kak script provides with terminal splitting using the
    tmux multiplexer

  • the ctags.kak script provides with symbol lookups using the
    readtags utility provided by some ctags implementations

Naming convention

All options and commands declared in a Kakoune script have to be
prefixed with the name of the script, or a one word description of the
purpose of the script.


  • in tmux.kak: command tmux-new-window

  • in comment.kak: option comment_line

The following conventions apply as well:

  • options: if a separator is needed to separate a multiple word
    option name, an underscore should be used to allow shell scopes to
    use them

  • commands: if a separator is needed, a hyphen is usually used to
    differentiate a command name from an option’s


Non-hidden commands and options should always be declared with a
documentation string, so that their purpose is clearly described
whenever completed upon interactively from the prompt.

POSIX shell

Shell expansions are a useful tool to interact with an external utility,
and the shell code that they contain should be as portable as possible.
As such, scripts that rely on those expansions have to be implemented
with POSIX in mind, most shells follow this standard nowadays which
somewhat guarantees that the script will be portable across the most
common systems.

Common shell patterns

Printing variables

In order to print a string that contains a variable expansion, prefer
printf to echo, as the latter is implementation defined and may
interpret some characters differently depending on the shell (e.g.
flags, backslashes).

        printf %s\\n "${var}"
        printf "value: %s\\n" "${var}"

The following won’t cause any issues, as the string to print doesn’t
contain ambiguous characters:

        echo "set global scrolloff 999,0"

Variable base name

Replace $(basename "${var}") with "${var##*/}".


The [[ keyword is provided by bash, and should be replaced with [.

Standard error redirection

Redirecting both standard and error streams is simplified in the bash
shell with the &> operator, however this syntax is not portable and
has to be replaced with the following: >/dev/null 2>&1.

Regular expression

The bash shell provides with a [[ keyword that supports the =~
operator to match a regular expression against a variable. This
functionality can be implemented with the expr utility:

  • expr "${var}" : '[a-z]*' >/dev/null returns successfully when the
    variable is empty or only contains lowercase characters, otherwise a
    non-zero exit code is returned

  • expr "${var}" : '\([a-z]*\)' prints the variable when empty or
    only contains lowercase characters

Note that the regular expression matches the whole string, using the ^
and $ anchors is an undefined behavior.

Running a process in the background

In order to get a process running in the background without having it
quit when the shell scope that spawns it terminates, use the following

        } </dev/null >/dev/null 2>&1 &
1 Like

I don’t know if this belong here but could you give an example on how to use kak -p ?
I tried
kak -s myses
and in another terminal
echo 'ihello' | kak -p myses
and also
echo "execute-keys 'ihello' " | kak -p myses
Expecting to write “hello” in my buffer but nothing happend.
Am I using it right ?

Ok I got it.
You need to use the -client option like so:

echo "execute-keys -client $client_id 'ihello' " | kak -p myses

If you just want to modify the buffer and don’t want to mess with the client’s state (like whether it’s in insert mode, etc.) you could also do something like:

echo "execute-keys -buffer /path/to/file 'ihello'" | kak -p myses

The documentation describes -buffer as “run in a disposable context for each given buffer”, where “context” includes things like “are we in insert mode” and “where is the cursor”.