Simple YAML-based task runner written in Go.

• Not super slow
• Not super obscure
• No dependencies
• Variables
• Simple

Via binary releases.

Via go-get:

$ go get github.com/tj/robo

Command-line usage.

Output tasks:

$ robo

  aws – amazon web services cli
  circle.open – open the repo in circle ci
  events – send data to the "events" topic
  push – push image from the current directory

Output task help:

$ robo help events

  Usage:

    events [project-id] [rate]

  Description:

    send data to the "events" topic

  Examples:

    Send 25 events a second to gy2d
    $ robo events gy2d 25

Regardless of task type (shell, exec, script) any additional arguments given will be passed.

For example suppose you have the following task:

aws:
  exec: ssh tools aws

You may then interact with the AWS cli as you would normally:

$ robo aws help
$ robo aws ec2 describe-instances

Task configuration.

The most basic task simply runs a shell command:

hello:
  summary: some task
  command: echo world

You may also define multi-line commands with YAML's |:

hello:
  summary: some task
  command: |
    echo hello
    echo world

Commands are executed via sh -c, thus you may use shell features and positional variables, for example:

hello:
  command: echo "Hello ${1:-there}"

Yields:

$ robo hello
Hello there

$ robo hello Tobi
Hello there Tobi

The exec alternative lets you replace the robo image without the fork & shell, however note that shell features are not available (pipes, redirection, etc).

hello:
  summary: some task
  exec: echo hello

Any arguments given are simply appended.

Shell scripts may be used instead of inline commands:

hello:
  summary: some task
  script: path/to/script.sh

If the script is executable, it is invoked directly, this allows you to use #!:

$ echo -e '#!/usr/bin/env ruby\nputs "yo"' > yo.rb
$ chmod +x yo.rb
$ cat > robo.yml
yo:
  summary: yo from rb
  script: yo.rb
^C
$ robo yo
yo

Script paths are relative to the config file, not the working directory.

Tasks may optionally specify usage parameters, which display upon help output:

events:
  summary: send data to the "events" topic
  exec: docker run -it events
  usage: "[project-id] [rate]"

Tasks may optionally specify any number of example commands, which display upon help output:

events:
  summary: send data to the "events" topic
  exec: docker run -it events
  usage: "[project-id] [rate]"
  examples:
    - description: Send 25 events a second to gy2d
      command: robo events gy2d 25

Robo supports variables via the text/template package. All you have to do is define a map of variables and use {{ }} to refer to them.

Here's an example:

stage:
  summary: Run commands against stage.
  exec: ssh {{.hosts.stage}} -t robo

prod:
  summary: Run commands against prod.
  exec: ssh {{.hosts.prod}} -t robo

variables:
  hosts:
    prod: bastion-prod
    stage: bastion-stage

Along with your own custom variables, robo defines the following variables:

$ robo variables

    robo.file: /Users/amir/dev/src/github.com/tj/robo/robo.yml
    robo.path: /Users/amir/dev/src/github.com/tj/robo

    user.home: /Users/amir
    user.name: Amir Abushareb
    user.username: amir

Tasks may define env key with an array of environment variables, this allows you to re-use robo configuration files, for example:

// aws.yml
dev:
  summary: AWS commands in dev environment
  exec: aws
  env: ["AWS_PROFILE=eng-dev"]

stage:
  summary: AWS commands in stage environment
  exec: aws
  env: ["AWS_PROFILE=eng-stage"]

prod:
  summary: AWS commands in prod environment
  exec: aws
  env: ["AWS_PROFILE=eng-prod"]

You can also override environment variables:

$ cat > robo.yml
home:
  summary: overrides $HOME
  exec: echo $HOME
  env: ["HOME=/tmp"]
^C
$ robo home // => /tmp

Variables can also be used to set env vars.

$ cat > robo.yml
aws-stage:
  summary: AWS stage
  exec: aws
  env: ["AWS_PROFILE={{.aws.profile}}"]
variables:
  aws:
    profile: eng-stage
^C
$ robo aws-stage ...

Note that you cannot use shell featurs in the environment key.

Task list and help output may be re-configured, for example if you prefer to view usage information instead of the summary:

templates:
  list: |
    {{range .Tasks}}  {{cyan .Name}} – {{.Usage}}
    {{end}}

Or perhaps something more verbose:

templates:
  list: |
    {{range .Tasks}}
      name: {{cyan .Name}}
      summary: {{.Summary}}
      usage: {{.Usage}}
    {{end}}

By default ./robo.yml is loaded, however if you want global tasks you can simply alias to something like:

alias segment='robo --config ~/.robo.yml'

You can easily use Robo to chain Robo, which is useful for multi-environment setups. For example:

prod:
  summary: production tasks
  exec: robo --config production.yml

stage:
  summary: stage tasks
  exec: robo --config stage.yml

Or on remote boxes:

prod:
  summary: production tasks
  exec: ssh prod-tools -t robo --config production.yml

stage:
  summary: stage tasks
  exec: ssh stage-tools -t robo --config stage.yml

You can also use robo's builtin variables robo.path, for example if you put all robofiles in together:

├── dev.yml
├── prod.yml
├── root.yml
└── stage.yml

And you would like to call dev, prod and stage from root:

dev:
  summary: Development commands
  exec: robo --config {{ .robo.path }}/dev.yml

stage:
  ...

You can compose multiple commands into a single command by utilizing robo's built-in robo.file variable:

one:
  summary: echo one
  command: echo one

two:
  summary: echo two
  command: echo two

all:
  summary: echo one two
  command: |
    robo -c {{ .robo.file }} one
    robo -c {{ .robo.file }} two

$ robo all
one
two

We generally use Makefiles for project specific tasks, however the discoverability of global tasks within a large team is difficult unless there's good support for self-documentation, which Make is bad at.

I'm aware of the million other solutions (Sake, Thor, etc) but I just wanted something fast without dependencies.

MIT