Skip to content


milpa help docs milpa quick-guide

Getting started with milpa is a matter of following these steps:

  1. Installing it,
  2. installing milpa's autocomplete script to your shell, possibly restarting your session,
  3. creating a new command, and editing it's spec
  4. running the command to test it out

You'll be creating some files during this guide, so make sure to create a folder wherever you like. Feel free to use an existing project folder if you already have a script for it in mind!


Installing milpa

Let's begin by running the following command to install the latest version of milpa:

curl -L | bash -

If homebrew is available, you can also install milpa with:

 brew install unRob/formulas/milpa

If all goes well, you'll see output that ends with something like:

🌽 Installed milpa version 3.1.4 🌽
Run 'milpa itself install-autocomplete' to install shell completions

You may run milpa --version to test milpa got installed correctly.

Installing autocomplete

One of the best things about milpa is it's built-in support for shell completion. This is what happens after you type milpa at a terminal prompt and press the tab key.

milpa will offer suggestions on commands, arguments and options based on your commands' spec. We'll install the required scripts by running:

milpa itself install-autocomplete

After this is done, you'll likely need to reload your shell or open a new session/tab. Verifying it's installed correctly can be done like so:

# type milpa, followed by a space, then press <TAB>
# you'll see this temporary output (on a zsh prompt, for this example):
# help      -- Display usage information on any command
# itself    -- itself subcommands

Creating a new command

Let's assume we want to build a script that every new hire to our programming coop runs to setup their system. Once milpa is installed, we'll use milpa itself create to create a new command, composed of a bash script and a YAML spec:

# Start by heading to your project's git repo
cd ~/src/my-teams-app

# milpa helps you get started by running `milpa itself create` to create a new command
# You can see how to use this command by running:
milpa itself create --help

# Now we've seen the help page for the `itself create` command,
# let's go ahead and create our "developer-setup" command:
milpa itself create developer-setup \
  --summary "Bootstrap a developer's machine" \
  --description "Installs all the fun stuff required for doing your dev job."

A couple of new files will be created, we verify with:

ls -lah .milpa/commands
# you'll see output like:
# -rw-r--r--  1 ada  staff  3141 Mar  14 15:09
# -rw-r--r--  1 ada  staff   314 Mar  14 15:09 developer-setup.yaml

Actually writing a command

Finally, we'll begin working on our command by editing our spec:

# open the yaml spec in your editor
"$EDITOR" .milpa/commands/developer-setup.yaml

Let's modify our spec at .milpa/commands/developer-setup.yaml to look like this:

# .milpa/commands/developer-setup.yaml
# We'll set a one-line `summary` and a more elaborated `description`
summary: Bootstrap a developer's machine
description: |
  Installs all the fun stuff required for doing your dev job.  

# We'll add a --kind option to specify which kind of work the user will be doing
    default: fullstack
    values: [fullstack, data]
    description: What kind of work will you be doing primarily?

Now, we should edit our script at .milpa/commands/ to do fun stuff:

#!/usr/bin/env bash
# .milpa/commands/

@milpa.log info "πŸš€ getting your computer ready for your ${MILPA_OPT_KIND} exploits πŸš€"

@milpa.log info "installing some packages..."
# then it's really up to you, here's some ideas:
# xcode-select --install
# brew bundle install --no-lock --file some.Brewfile
@milpa.log info "configuring credentials sources"
# aws configure ...
# gcloud auth ...
# op login ...
@milpa.log info "pleasing the compliance overlords, overladies, and overfolks"
# defaults write ScheduleFrequency -int 1
# defaults write askForPassword -int 1
# defaults write askForPasswordDelay -int 0

@milpa log complete "Your system is ready to roll, $(whoami)!"

Once we're happy with our developer setup script, we can check milpa recognizes the command and can parse it's spec:

milpa itself doctor
# output will look like:
# MILPA_ROOT is: /usr/local/lib/milpa
# /usr/local/lib/milpa
# /home/ada/src/my-teams-app
# Runnable commands:
# -----------
# βœ… developer-setup β€” /home/ada/my-teams-app/.milpa/commands/
# -----------
# βœ… itself create β€” /usr/local/lib/milpa/.milpa/commands/itself/
# -----------
# .. and so on
# -----------

If you don't see a checkmark next to your command's name, you'll get some feedback on what went wrong. Remember you can always run milpa help docs milpa command spec to see all the documentation on the command spec format.

Running your new command

Once everything looks good, we should be ready to test our command out!

milpa developer-setup
# you'll see output like:
# [info:developer-setup] πŸš€ getting your computer ready for your fullstack exploits πŸš€
# ...
# [info:developer-setup] βœ… Your system is ready to roll, Ada!

Well done, $(whoami)! You've created and ran your first command. milpa comes with documentation on these commands and more, run milpa help docs milpa command to find out more.

Finally, if you want to see the rest of the owlβ€”that is, an actual implementation of this sort of setup scriptβ€”then head over to unRob/dotfiles where you'll a practical example of how I setup my personal and work computers.

Further reference

As you continue working on your scripts, make sure to check out these reference docs: