Operating system shells have long had the ability to execute a sequence of commands saved in a text file. These script files make long sequences of commands easier to repeatedly execute. cmd2 supports two similar mechanisms: command scripts and python scripts.

Command Scripts

A command script contains a sequence of commands typed at the the prompt of a cmd2 based application. Unlike operating system shell scripts, command scripts can’t contain logic or loops.

Creating Command Scripts

Command scripts can be created in several ways:

  • creating a text file using any method of your choice
  • using the built-in edit command to create or edit an existing text file
  • saving previously entered commands to a script file using history -s

If you create create a text file from scratch, just include one command per line, exactly as you would type it inside a cmd2 application.

Running Command Scripts

Command script files can be executed using the built-in run_script command or the @ shortcut (if your application is using the default shortcuts). Both ASCII and UTF-8 encoded unicode text files are supported. The run_script command supports tab completion of file system paths. There is a variant _relative_run_script command or @@ shortcut (if using the default shortcuts) for use within a script which uses paths relative to the first script.


Any command line input where the first non-whitespace character is a # will be treated as a comment. This means any # character appearing later in the command will be treated as a literal. The same applies to a # in the middle of a multiline command, even if it is the first character on a line.

Comments are useful in scripts, but would be pointless within an interactive session.

(Cmd) # this is a comment
(Cmd) command # this is not a comment

Python Scripts

If you require logic flow, loops, branching, or other advanced features, you can write a python script which executes in the context of your cmd2 app. This script is run using the run_pyscript command. Here’s a simple example that uses the arg_printer script:

(Cmd) run_pyscript examples/scripts/ foo bar 'baz 23'
Running Python script '' which was called with 3 arguments
arg 1: 'foo'
arg 2: 'bar'
arg 3: 'baz 23'

run_pyscript supports tab completion of file system paths, and as shown above it has the ability to pass command-line arguments to the scripts invoked.

Developing a CMD2 API

If you as an app designer have not explicitly disabled the run_pyscript command it must be assumed that your application is structured for use in higher level python scripting. The following sections are meant as guidelines and highlight possible pitfalls with both production and consumption of API functionality. For clarity when speaking of “scripter” we are referring to those writing scripts to be run by pyscript and “designer” as the CMD2 application author.


Without any work on the part of the designer, a scripter can take advantage of piecing together workflows using simple app calls. The result of a run_pyscript app call yields a CommandResult object exposing four members: Stdout, Stderr, Stop, and Data.

Stdout and Stderr are fairly straightforward representations of normal data streams and accurately reflect what is seen by the user during normal cmd2 interaction. Stop contains information about how the invoked command has ended its lifecycle. Lastly Data contains any information the designer sets via self.last_result or self._cmd.last_result if called from inside a CommandSet.

Python scripts executed with run_pyscript can run cmd2 application commands by using the syntax:

app(‘command args’)


  • app is a configurable name which can be changed by setting the cmd2.Cmd.py_bridge_name attribute
  • command and args are entered exactly like they would be entered by a user of your application.

Using fstrings tends to be the most straight forward and easily readable way to provide parameters.:

first = 'first'
second = 'second'

app(f'command {first} -t {second})

See python_scripting example and associated conditional script for more information.

Design principles

If the cmd2 application follows the unix_design_philosophy a scriptor will have the most flexibility to piece together workflows using different commands. If the designers’ application is more complete and less likely to be augmented in the future a scripter may opt for simple serial scripts with little control flow. In either case, choices made by the designer will have effects on scripters.

The following diagram illustrates the different boundaries to keep in mind.

|                                             |
| Py scripts                                  |
|                                             |
| +-----------------------------------------+ |
| |  CMD2 Application                       | |
| |                                         | |
| | +-------------------------------------+ | |
| | |   Class Library                     | | |
| | | +------+ +------+ +------+ +------+ | | |
| | | |      | |      | |      | |      | | | |
| | | |  C   | |  C   | |  C   | |  C   | | | |
| | | |      | |      | |      | |      | | | |
| | | +------+ +------+ +------+ +------+ | | |
| | |                                     | | |
| | +-------------------------------------+ | |
| |                                         | |
| +-----------------------------------------+ |
|                                             |


As a designer it is preferable to design from the inside to out. Your code will be infinitely far easier to unit test than at the higher level. While there are regression testing extensions for cmd2 UnitTesting will always be faster for development.


It is bad design or a high level py_script to know about let alone access low level class libraries of an application. Resist this urge at all costs, unless it’s necessary.

Developing a Basic API

CMD2 out of the box allows scripters to take advantage of all exposed do_* commands. As a scripter one can easily interact with the application via stdout and stderr.

As a baseline lets start off with the familiar FirstApp

#!/usr/bin/env python
"""A simple cmd2 application."""
import cmd2

class FirstApp(cmd2.Cmd):
    """A simple cmd2 application."""
    def __init__(self):
      shortcuts = cmd2.DEFAULT_SHORTCUTS
      shortcuts.update({'&': 'speak'})

      # Make maxrepeats settable at runtime
      self.maxrepeats = 3
      self.add_settable(cmd2.Settable('maxrepeats', int, 'max repetitions for speak command', self))

    speak_parser = cmd2.Cmd2ArgumentParser()
    speak_parser.add_argument('-p', '--piglatin', action='store_true', help='atinLay')
    speak_parser.add_argument('-s', '--shout', action='store_true', help='N00B EMULATION MODE')
    speak_parser.add_argument('-r', '--repeat', type=int, help='output [n] times')
    speak_parser.add_argument('words', nargs='+', help='words to say')

    def do_speak(self, args):
        """Repeats what you tell me to."""
        words = []
        for word in args.words:
            if args.piglatin:
                word = '%s%say' % (word[1:], word[0])
            if args.shout:
                word = word.upper()
        repetitions = args.repeat or 1
        for _ in range(min(repetitions, self.maxrepeats)):
            # .poutput handles newlines, and accommodates output redirection too
            self.poutput(' '.join(words))

if __name__ == '__main__':
    import sys
    c = FirstApp()

Lets start off on the wrong foot:

SyntaxError: unexpected EOF while parsing
(Cmd) run_pyscript
  File "<string>", line 2
SyntaxError: unexpected EOF while parsing

cmd2 pyscripts require valid python code as a first step.


It is a common misconception that all application exceptions will “bubble” up from below. Unfortunately or fortunately this is not the case. CMD2 sinkholes all application exceptions and there are no means to handle them.

When executing the speak command without parameters you see the following error:

(Cmd) speak
Usage: speak [-h] [-p] [-s] [-r REPEAT] words [...]
Error: the following arguments are required: words

Even though this is a fully qualified CMD2 error the py_script must look for this error and perform error checking.:

(Cmd) run_pyscript

You should notice that no error message is printed. Let’s utilize the CommandResult object to inspect the actual returned data.:

result = app('speak')
(Cmd) run_pyscript
CommandResult(stdout='', stderr='Usage: speak [-h] [-p] [-s] [-r REPEAT] words [...]\nError: the following arguments are required: words\n\n', stop=False, data=None)

Now we can see that there has been an error. Let’s re write the script to perform error checking.:

result = app('speak')

if not result:
(Cmd) run_pyscript
Something went wrong

In python development is good practice to fail and exit quickly after user input.:

import sys

result = app('speak TRUTH!!')

if not result:
    print("Something went wrong")

print("Continuing along..")
(Cmd) run_pyscript
Continuing along..

We changed the input to be a valid speak command but no output. Again we must inspect the CommandResult:

import sys

#Syntax error
result = app('speak TRUTH!!!')
if not result:
    print("Something went wrong")

(Cmd) run_pyscript

By just using stdout and stderr it is possible to string together commands with rudimentary control flow. In the next section we will show how to take advantage of cmd_result data.

Developing an Advanced API

Until now the application designer has paid little attention to scripters and their needs. Wouldn’t it be nice if while creating py_scripts one did not have to parse data from stdout? We can accomodate the weary scripter by adding one small line at the end of our do_* commands.

self.last_result = <value>

Adding the above line supercharges a cmd2 application and opens a new world of possibilities.


When setting results for a command function inside of a CommandSet use the private cmd instance:

self._cmd.last_result = <value>

In the following command example we return an array containing directory elements.:

dir_parser = cmd2.Cmd2ArgumentParser()
dir_parser.add_argument('-l', '--long', action='store_true', help="display in long format with one item per line")

@cmd2.with_argparser(dir_parser, with_unknown_args=True)
def do_dir(self, args, unknown):
    """List contents of current directory."""
    # No arguments for this command
    if unknown:
        self.perror("dir does not take any positional arguments:")

    # Get the contents as a list
    contents = os.listdir(self.cwd)

    for f in contents:

    self.last_result = contents

The following script retrieves the array contents.:

result = app('dir')


Cmd) run_pyscript
['.venv', '', '']

As a rule of thumb it is safer for the designer to return simple scalar types as command results instead of complex objects. If there is benefit in providing class objects designers should choose immutable over mutable types and never provide direct access to class members as this could potentially lead to violation of the open_closed_principle.

When possible, a dataclass is a lightweight solution perfectly suited for data manipulation. Lets dive into an example.

The following fictitional application has two commands: build and status. We can pretend that the build action happens somewhere else in the world at an REST API endpoint and has significant computational cost. The status command for all intents and purposes will only show the current status of a build task. The application has provided all that is needed for a user to start a build and then determine it’s status. The problem however is that with a long running process the user may want to wait for it to finish. A designer may be tempted to create a command to start a build and then poll for status until finished but this scenario is better solved as an extensible script.

#!/usr/bin/env python
"""A simple cmd2 application."""
import sys
from dataclasses import dataclass
from random import choice, randint
from typing import Optional

import cmd2
from cmd2.parsing import Statement

class BuildStatus:
    id: int
    name: str
    status: str

class FirstApp(cmd2.Cmd):
    """A simple cmd2 application."""

    def __init__(self):
        self._status_cache = dict()

    def _start_build(self, name: str) -> BuildStatus:
        return BuildStatus(randint(10, 100), name, "Started")

    def _get_status(self, name: str) -> Optional[BuildStatus]:

        status = self._status_cache.get(name)

        status_types = ["canceled", "restarted", "error", "finished"]

        if status.status != "finished":
            status = BuildStatus(,, choice(status_types))
            self._status_cache[name] = status

        return status

    build_parser = cmd2.Cmd2ArgumentParser()
    build_parser.add_argument("name", help="Name of build to start")

    def do_build(self, args: Statement):
        """Executes a long running process at an API endpoint"""
        status = self._start_build(
        self._status_cache[] = status

            f"Build {} successfuly stared with id : {}"
        self.last_result = status

    status_parser = cmd2.Cmd2ArgumentParser()
    status_parser.add_argument("name", help="Name of build determine status of")

    def do_status(self, args: Statement):
        """Shows the current status of a build"""

        status = self._get_status(

        self.poutput(f"Status for Build: {} \n {status.status}")
        self.last_result = status

if __name__ == "__main__":
    import sys

    c = FirstApp()

The below is a possible solution via pyscript:

import sys
import time

# start build
result = app('build tower')

# If there was an error then quit now
if not result:
    print('Build failed')

# This is a BuildStatus dataclass object
build =

print(f"Build {} : {build.status}")

# Poll status (it would be wise to NOT hang here)
while True:

    # Perform status check
    result = app('status tower')

    #error checking
    if not result:
        print("Unable to determin status")

    build_status =

    # If the status shows complete then we are done
    if build_status.status in ['finished', 'canceled']:
        print(f"Build {} has completed")

    print(f"Current Status: {build_status.status}")