Shell Task Configuration (shell)

Run shell commands

The shell task type allows you to run shell commands.

Task Properties

Property	Type	Description
task	"shell", required	Specifies the task type, which should be set to "shell" for this task.
name	string	An optional name for the task. If provided, the task state will be saved as a variable. Visit Task and Action States page to learn more.
label	string	An optional label or description for the task.
when	object	Visit Conditional Tasks and Actions page to learn how to execute task conditionally.
actions	Array<Action>, required	An array of action items that define the modifications to be made in the file.

Action Properties

Property	Type	Description
name	string	An optional name for the task. If provided, the task state will be saved as a variable. Visit Task and Action States page to learn more.
when	object	Visit Conditional Tasks and Actions page to learn how to execute action conditionally.
command	string, required	The process name to spawn. Command can have simple args, parser supports quotes. For more complex args, use args property.
args	Array<string>	Array of args to be past directly to process.
cwd	string	Working directory to run shell. Must be relative to the project root. Default is project root.

Example

Here's an example of how to use shell task in a configuration file:

steps:
  - task: shell
    label: Embedding assets
    actions:
      - command: npx react-native-asset
      - command: some-command --some-flag "some literal args"
      - command: other-command
        args:
          - --some-flag
          - "some \"complex\" args"

tip

Specify name field for this action to expose the name.output variable which will hold the output of the process.

Example:

  # run command 
  - command: npx react-native-asset
    name: cmd_asset # Give it a name

  # you can run another command if previous was success and output contains some value
  - when:
      cmd_asset: done
      cmd_asset.output: # use the name here
        $regex: somevalue
    command: some-other-command

warning

shell commands that are used in package integrations will always ask permission from user before execution. User may choose to skip the execution, so don't rely on it too much.

It does not ask permission when used in upgrade.yml.