cmd2 is designed to make it easy for you to create new commands. These commmands form the backbone of your application. If you started writing your application using cmd, all the commands you have built will work when you move to cmd2. However, there are many more capabilities available in cmd2 which you can take advantage of to add more robust features to your commands, and which makes your commands easier to write. Before we get to all the good stuff, let’s briefly discuss how to create a new command in your application.

Basic Commands

The simplest cmd2 application looks like this:

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

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

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

This application subclasses cmd2.Cmd but has no code of it’s own, so all functionality (and there’s quite a bit) is inherited. Lets create a simple command in this application called echo which outputs any arguments given to it. Add this method to the class:

def do_echo(self, line):

When you type input into the cmd2 prompt, the first space delimited word is treated as the command name. cmd2 looks for a method called do_commandname. If it exists, it calls the method, passing the rest of the user input as the first argument. If it doesn’t exist cmd2 prints an error message. As a result of this behavior, the only thing you have to do to create a new command is to define a new method in the class with the appropriate name. This is exactly how you would create a command using the cmd module which is part of the python standard library.


See Generating Output if you are unfamiliar with the poutput() method.


A command is passed one argument: a string which contains all the rest of the user input. However, in cmd2 this string is actually a Statement object, which is a subclass of str to retain backwards compatibility.

cmd2 has a much more sophsticated parsing engine than what’s included in the cmd module. This parsing handles:

  • quoted arguments
  • output redirection and piping
  • multi-line commands
  • shortcut, macro, and alias expansion

In addition to parsing all of these elements from the user input, cmd2 also has code to make all of these items work; it’s almost transparent to you and to the commands you write in your own application. However, by passing your command the Statement object instead of just a plain string, you can get visibility into what cmd2 has done with the user input before your command got it. You can also avoid writing a bunch of parsing code, because cmd2 gives you access to what it has already parsed.

A Statement object is a subclass of str that contains the following attributes:

Name of the command called. You already know this because of the method cmd2 called, but it can sometimes be nice to have it in a string, i.e. if you want your error messages to contain the command name.
A string containing the arguments to the command with output redirection or piping to shell commands removed. It turns out that the “string” value of the Statement object has all the output redirection and piping clauses removed as well. Quotes remain in the string.
A string of just the command and the arguments, with output redirection or piping to shell commands removed.
A list of arguments a-la sys.argv, including the command as argv[0] and the subsequent arguments as additional items in the list. Quotes around arguments will be stripped as will any output redirection or piping portions of the command.
Full input exactly as typed by the user.
Character used to end a multiline command. You can configure multiple termination characters, and this attribute will tell you which one the user typed.

For many simple commands, like the echo command above, you can ignore the Statement object and all of it’s attributes and just use the passed value as a string. You might choose to use the argv attribute to do more sophisticated argument processing. Before you go too far down that path, you should check out the Argument Processing functionality included with cmd2.

Return Values

Most commands should return nothing (either by omitting a return statement, or by return None. This indicates that your command is finished (with or without errors), and that cmd2 should prompt the user for more input.

If you return True from a command method, that indicates to cmd2 that it should stop prompting for user input and cleanly exit. cmd2 already includes a quit command, but if you wanted to make another one called finish you could:

def do_finish(self, line):
    """Exit the application"""
    return True

Exit Codes

cmd2 has basic infrastructure to support sh/ksh/csh/bash type exit codes. The cmd2.Cmd object sets an exit_code attribute to zero when it is instantiated. The value of this attribute is returned from the cmdloop() call. Therefore, if you don’t do anything with this attribute in your code, cmdloop() will (almost) always return zero. There are a few built-in cmd2 commands which set exit_code to 1 if an error occurs.

You can use this capability to easily return your own values to the operating system shell:

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

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

def do_bail(self, line):
    """Exit the application"""
    self.perror("fatal error, exiting")
    self.exit_code = 2
    return true

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

If the app was run from the bash operating system shell, then you would see the following interaction:

(Cmd) bail
fatal error, exiting
$ echo $?

Raising SystemExit(code) or calling sys.exit(code) in a command or hook function also sets self.exit_code and stops the program.

Exception Handling

You may choose to catch and handle any exceptions which occur in a command method. If the command method raises an exception, cmd2 will catch it and display it for you. The debug setting controls how the exception is displayed. If debug is false, which is the default, cmd2 will display the exception name and message. If debug is true, cmd2 will display a traceback, and then display the exception name and message.

There are a few exceptions which commands can raise that do not print as described above:

  • cmd2.exceptions.SkipPostcommandHooks - all postcommand hooks are skipped and no exception prints
  • cmd2.exceptions.Cmd2ArgparseError - behaves like SkipPostcommandHooks
  • SystemExit - stop will be set to True in an attempt to stop the command loop
  • KeyboardInterrupt - raised if running in a text script and stop isn’t already True to stop the script

All other BaseExceptions are not caught by cmd2 and will be raised

Disabling or Hiding Commands

See Disabling Commands for details of how to:

  • remove commands included in cmd2
  • hide commands from the help menu
  • disable and re-enable commands at runtime

Modular Commands and Loading/Unloading Commands

See Modular Commands for details of how to:

  • Define commands in separate CommandSet modules
  • Load or unload commands at runtime