Writing scaffolder templates

Published on May 16th, 2022

Overview

The Roadie Backstage scaffolder is a feature that allows you to define software templates to create new software projects, update existing ones or simply perform repeated tasks in a consistent manner.

Scaffolder templates are defined in YAML files and loaded into the Backstage catalog in the same way that other entities are loaded into Backstage. A template contains one or more steps which run sequentially during execution.

A Scaffolder template is then run on demand by the users of Backstage to execute the software template. Roadie will execute the software template in an ephemeral container that is destroyed after the execution completes.

You can find a step by step guide to adding templates in Roadie here

See the Backstage Scaffolder in action

Request a Roadie demo

Components of a Template

A Scaffolder template is a configurable process that will run one or more Scaffolder steps. The template will be run when a user visits the “Create Component” page in Backstage. https://<tenant-name>.roadie.so/create.

create-a-new-component

Templates are defined by a Backstage Entity YAML file with a Template kind and imported into the Backstage catalog. You can create multiple templates, each of which can perform a different set of steps. For example, you can have one template that creates a React application, and another that creates a serverless app.

Template YAML input forms can be tested at /tools using a live template preview viewer.

preview-template

Here is an example of a very basic Scaffolder template that prompts the user for a name, and then prints back the text “Hello, name!”

apiVersion: scaffolder.backstage.io/v1beta3
kind: Template
metadata:
  name: hello-world-template
  title: Hello World
  description: Says Hello to a specified name.
spec:
  owner: backstage/techdocs-core
  type: service

  parameters:
    - title: You are about to say hello to your first Backstage Template
      required:
        - name
      properties:
        name:
          type: string

  steps:
    - id: log-message
      name: Log Message
      action: debug:log
      input:
        message: 'Hello, ${{ parameters.name }}!'

Header Section

The header section is required for every template and contains information to configure the task and show details about the task on the “Create Component” page.

apiVersion: scaffolder.backstage.io/v1beta3
kind: Template
metadata:
  name: hello-world-template
  title: Hello World
  description: Says Hello to a specified name.
spec:
  owner: default/engineering
  type: service

apiVersion

This is a required field and should be set to scaffolder.backstage.io/v1beta3

kind

A Scaffolder template is also an Entity in Backstage. In order to configure this entity as a template you must set the kind to Template

metadata

The metadata field contains some data that appears on the template card that appears on the “Create Component” page.

spec

The spec field contains owner and type. Owner refers to the Backstage group or user that owns the Scaffolder task e.g. default/engineering. Type refers to the type of template. It can be set to anything and appears on the scaffolder template card in the “Create Component” page.

parameters

The parameters property is a list of parameters that can be prompted from the user when they run a template. Each array element contains the configuration for a single page of items to be filled by the user running the template. The parameter pages must contain title, required and properties.

The parameters yaml is based on react-jsonschema-form. You can find the available syntax options there and examples here.

You can choose to break up the parameter prompting into form steps or collect all the parameters in one single step. Each parameter can be one of a few types: string, number, array or object.

Here is the most basic example:

parameters:
  properties:
    name:
      type: string

Validation

You can use react-jsonschema-form to perform validation on input fields using a regex or character counts.

parameters:
  properties:
    name:
        title: Simple text input
        type: string
        description: Description about input
        maxLength: 8
        pattern: '^([a-zA-Z][a-zA-Z0-9]*)(-[a-zA-Z0-9]+)*$'
        ui:autofocus: true
        ui:help: 'Hint: additional description...'

string

You may collect text data from the user by using the string type. Here is the most basic example. It will prompt the user for a name.

parameters:
  properties:
    name:
      type: string

Entity picker

You can prompt the user with a list of catalog entities using the ui:field: EntityPicker option as follows:

parameters:
  properties:
    entity:
      type: string
      ui:field: EntityPicker

Owned entity picker

Alternatively if you would like the user to only select entities that they already own, you might want to use the OwnedEntityPicker.

parameters:
  properties:
    ownedEntity:
      type: string
      ui:field: OwnedEntityPicker

Entity name picker

If you would like a little validation when the user enters an Entity name, you can use the EntityNamePicker. It will prevent the user from entering an entity name that is not an acceptable entity name.

parameters:
  properties:
    ownedEntity:
      type: string
      ui:field: EntityNamePicker

Repository picker

The respository picker can allow the user to select the name and location of a new repository. The picker restricts the target location of the repository to make it a little easier for the user to select a target location.

The following example, will only allow the user to enter a new repository name targeting the GitHub using the AcmeInc organization.

parameters:
  properties:
    repoUrl:
      type: string
      ui:field: RepoUrlPicker
      ui:options:
        allowedHosts:
          - github.com
        allowedOwners:
          - AcmeInc

The RepoUrlPicker uses the allowedHosts to decide how to build the repo url output value. If you use bitbucket.org it will output a valid repo url for Bitbucket.

parameters:
  properties:
    repoUrl:
      type: string
      ui:field: RepoUrlPicker
      ui:options:
        allowedHosts:
          - bitbucket.org

Owner picker

The owner picker, allows the user to select a user / group in the Backstage catalog. e.g.

parameters:
  properties:
    owner:
      type: string
      ui:field: OwnerPicker

This returns a variable in the format group:<namespace>/<group-or-user-name>. You can extract the entity name using replace when you refer to the parameter like so: ${{ parameters.owner | replace(\"group:.*/\", \"\") }}

Picker from external API source

This custom scaffolder field, makes an API call to the Backstage backend and allows the result to be rendered to a list.

parameters:
  properties:
    custom:
      title: custom
      type: string
      description: Custom field from external API

      # Use `SelectFieldFromApi` to configure the select field for the entry.
      ui:field: SelectFieldFromApi

      ui:options:
        # The Path on the Backstage API and the parameters to fetch the data for the dropdown
        path: 'catalog/entity-facets'
        params:
          facet: 'kind'

        # This selects the array element from the API fetch response. It finds the array with the name kind
        # under the facets object
        arraySelector: 'facets.kind'

        # (Optional) This selects the field in the array to use for the value of each select item. If its not specified
        # it will use the value of the item directly.
        valueSelector: 'count'
        # (Optional) This selects the field in the array to use for the label of each select item.
        labelSelector: 'value'

number

You can allow the user to enter a number using the number type:

parameters:
  properties:
    size:
      type: number

object

The object allows the collection of more complex types of data from the user. It contains the properties option to add variables to the object as follows:

parameters:
  properties:
    person:
      type: object
      properties:
        name:
          type: string
        age:
          type: number

You may choose to make an object property to be mandatory using the required property.

parameters:
  properties:
    person:
      type: object
      required:
        - name
      properties:
        name:
          type: string
        age:
          type: number

array

You can prompt for an array of properties using the array option. The items option can be any type: array, object, string or number as you like.

parameters:
  properties:
    languages:
      type: array
      items:
        type: string

If you would like to prompt the user to add entity tags, you can use the ui:field: EntityTagPicker as shown below.

parameters:
  properties:
    entityTags:
      type: array
      ui:field: EntityTagsPicker

Outputs

Parameters can be retrieved later on by steps using parameter outputs. Here is an example of a parameter name being used by a debug:log step.

parameters:
  properties:
    name:
      type: string
steps:
  - id: log-message
    name: Log Message
    action: debug:log
    input:
      message: 'Hello, ${{ parameters.name }}!'

If you need to reference elements of an array parameter you can refer to them using the following syntax:

steps:
  - id: log-message
    name: Log Message
    action: debug:log
    input:
      message: 'Hello, ${{ parameters.names[0] }}!'

An object parameter values can be reference in the way you might expect.

steps:
  - id: log-message
    name: Log Message
    action: debug:log
    input:
      message: 'Hello, ${{ parameters.person.name }}!'

Common Options

If you would like to default the value of a field you can use the default option:

parameters:
  properties:
    name:
      type: string
      default: 'world!'

If you would like to prompt the users for a fixed list of options, you may use the enum option.

parameters:
  properties:
    size:
      type: number
      enum: [50, 100, 200]

You can display a more human description to a field value by using title and description

parameters:
  properties:
    name:
      type: string
      title: 'Name'
      description: 'Name to say hello to'

Form Steps

It might be jarring for your user to enter a lot of parameters one after another on the same page, especially if some properties require validation. As such Backstage have provided form steps.

You can make use of form steps using the following example.

parameters:
  - title: 'Fill in the Name'
    properties:
      name:
        type: string
  - title: 'Fill in the Age'
    properties:
      age:
        type: number

Previewing parameters

Template Preview, which is accessible via Tools > Template Preview provides a preview page for templates, where you can see a live preview of the template form. This is done in order to provide an easy way to preview scaffolder template form UIs without running your own local instance of the plugin or committing changes to the template.

parameters-preview

More Reading

You can read more about parameter configuration in the official backstage docs here.

steps

Steps define the actions that are taken by the scaffolder template when it is run as a task. The scaffolder initially creates a temporary directory referred to as the workspace, in which files are downloaded, generated, updated and pushed to some external system. Each step that is defined is run in order.

Parameters taken from the user earlier may be used in the action steps using the syntax ${{ parameters.name }}.

Step Outputs

You can refer to the output of a previous step using the following syntax:

${{ steps["publish-step-id"].output.repoContentsUrl }}

If the step id does not contain a special character you can also refer to it using the dot syntax.

${{ steps.publish.output.repoContentsUrl }}

Parameter Values

You can refer to the value of a parameter using the following syntax:

${{ parameters["name"] }}

If the parameter id does not contain a special character you can also refer to it using the dot syntax.

${{ parameters.name }}

fetch:plain

Downloads content and places it in the workspace.

steps:
  - action: fetch:plain
    id: fetch-plain
    name: Fetch plain
    input:
      url: ./plain

Optionally, if you would prefer the data to be downloaded to a subdirectory in the workspace you may specify the ‘targetPath’ input option.

steps:
  - action: fetch:plain
    id: fetch-plain
    name: Fetch plain
    input:
      url: ./plain
      targetPath: fetched-data

Outputs

The fetch:plain action does not output any data.

fetch:template

This downloads a directory containing templated files. It then renders all the template variables into the files and directory names and content, and places the result in the workspace.

steps:
  - action: fetch:template
    id: fetch-template
    name: Fetch template
    input:
      url: ./template
      values:
        name: ${{ parameters.name }}

The templated files themselves can contain references to the values in the following way ${{ values.name }}. It uses the nunjucks templating language. More details can be found here.

i.e. ./template/README.md

# ${{ values.name }} service

This is a service Readme example. Please update me.

Optionally, if you would prefer the data to be downloaded to a subdirectory in the workspace you may specify the ‘targetPath’ input option.

steps:
  - action: fetch:template
    id: fetch-template
    name: Fetch template
    input:
      url: ./template
      targetPath: fetched-data
      values:
        name: ${{ parameters.name }}

You can also choose to not template specific files downloaded by the task by using the copyWithoutRender option. It may use file paths or globs.

steps:
  - action: fetch:template
    id: fetch-template
    name: Fetch template
    input:
      url: ./template
      copyWithoutRender:
        - README.md
        - src/**.ts
      values:
        name: ${{ parameters.name }}

If you would like to limit the templating to very specific files, you can optionally add the .njk extension to the files and use the templateFileExtension option.

steps:
  - action: fetch:template
    id: fetch-template
    name: Fetch template
    input:
      url: ./template
      templateFileExtension: true
      values:
        name: ${{ parameters.name }}

Outputs

The fetch:template action does not output any data.

publish:github

This action creates a new GitHub repository and publishes the files in the workspace directory to the repository. There is one mandatory parameter repoUrl. The repo url picker described in the string parameter description above.

The repoUrl must be in the format github.com?repo=<reponame>&owner=<owner org>

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'

By default it will create a repository with a master branch. If you prefer to use main you can do the following:

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      defaultBranch: main

The access input parameter adds an admin collaborator to the repository. It can be a reference to a GitHub user or a team in GitHub.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      access: AcmeInc/engineering

You can protect the default branch from being pushed to directly by using protectDefaultBranch if your repository is part of a Github Pro account.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      protectDefaultBranch: true

You can enable code owner reviews using the requireCodeOwnerReviews option:

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      requireCodeOwnerReviews: true

The repoVisibility option allows the repository to be made public. By default it will be a private repository.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      repoVisibility: 'public'

To cause merges to delete the source branch, you can enable the deleteBranchOnMerge setting.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      deleteBranchOnMerge: true

If you want to disable merge commits, squash merge and rebase merge you can do that with the settings allowMergeCommit, allowSquashMerge and allowRebaseMerge. By default, these are enabled.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      allowMergeCommit: false
      allowSquashMerge: false
      allowRebaseMerge: false

By default the repository will be populated with the files contained in the workspace directory. If you need to use a subdirectory, you can use the sourcePath option.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      sourcePatch: './repoRoot'

Collaborators can be added to the repository using the collaborators option. It takes an array of username or team and access. username is the GitHub username to allow collaboration.

The access option gives the user specific type of permissions. The options are pull, push, admin, maintain or triage. these equate to:

  • pull (read)
  • push (write)
  • triage (triage)
  • admin (admin)
  • maintain (maintain - only for public repos)

The team value should be the Github team slug and should not include the org-name as a prefix.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      collaborators:
        - user: user1
          access: admin
        - team: github-team-name
          access: pull

The topics allows adding topics to the created repository when its created.

steps:
  - action: publish:github
    id: publish-repository
    name: Publish Repository to Github
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      topics:
        - java
        - ruby

Outputs

The publish:github action produces two step outputs.

Name Description
remoteUrl Url for the newly created repository
repoContentsUrl Url that shows the contents of the repository

These outputs can be retrieved by a subsequent step using:

steps:
  - id: log-message
    name: Log Message
    action: debug:log
    input:
      message: 'RemoteURL: ${{ steps["publish-repository"].output.remoteUrl }}, ${{ steps["publish-repository"].output.repoContentsUrl }}!'

publish:github:pull-request

This action creates a pull request against a pre-existing repository using the files contained in the workspace directory. The most basic example is:

steps:
  - action: publish:github:pull-request
    id: create-pull-request
    name: Create a pull request
    input:
      repoUrl: 'github.com?repo=reponame&owner=AcmeInc'
      branchName: ticketNumber-123
      title: 'Make some changes to the files'
      description: 'This pull request makes changes to the files in the reponame repository in the AcmeInc organization'

If the updated code is contained in a subdirectory to the workspace directory, you can use the sourcePath to select it. If the files you want to target to update are in a subdirectory of the repository you can use the targetPath option.

steps:
  - action: publish:github:pull-request
    id: create-pull-request
    name: Create a pull request
    input:
      repoUrl: 'github.com?repo=reponame&owner=AcmeInc'
      branchName: ticketNumber-123
      title: 'Make some changes to the files'
      description: 'This pull request makes changes to the files in the reponame repository in the AcmeInc organization'
      sourcePath: ./subdirectory
      targetPath: ./subdirectory

You can use the user that runs the scaffolder template to open the PR rather than opening it on behalf of the Roadie Github App by specifying the token field. The token can be injected by the RepoUrlPicker parameter as documented here

parameters:
  - title: Choose a location
    required:
      - repoUrl
    properties:
      repoUrl:
        title: Repository Location
        type: string
        ui:field: RepoUrlPicker
        ui:options:
          # Here's the option you can pass to the RepoUrlPicker
          requestUserCredentials:
            secretsKey: USER_OAUTH_TOKEN
            additionalScopes:
              github:
                - workflow
          allowedHosts:
            - github.com
steps:
  - action: publish:github:pull-request
    id: create-pull-request
    name: Create a pull request
    input:
      repoUrl: 'github.com?repo=reponame&owner=AcmeInc'
      branchName: ticketNumber-123
      title: 'Make some changes to the files'
      description: 'This pull request makes changes to the files in the reponame repository in the AcmeInc organization'
      # here's where the secret can be used
      token: ${{ secrets.USER_OAUTH_TOKEN }}

NB: The branch you use for the pull request must be a new branch for the repo.

Outputs

The publish:github:pull-request action produces two outputs.

Name Description
remoteUrl Url to the new pull request
pullRequestNumber Shows the number of the pull request

They can be accessed in subsequent steps as follows:

steps:
  - id: log-message
    name: Log Message
    action: debug:log
    input:
      message: 'RemoteURL: ${{ steps["create-pull-request.output.remoteUrl }}, ${{ steps["create-pull-request"].output.pullRequestNumber }}!'

publish:bitbucket

This action creates a new Bitbucket repository and publishes the files in the workspace directory to the repository. There is one mandatory parameter repoUrl. The repo url picker described in the string parameter description above.

The repoUrl must be in the format bitbucket.org?repo=<project name>&workspace=<workspace name>&project=<project name>

steps:
  - action: publish:bitbucket
    id: publish-repository
    name: Publish Repository to Bitbucket
    input:
      repoUrl: 'bitbucket.org?repo=newprojectname&workspace=workspacename&project=projectname'

You can optionally add a description to the new repository.

steps:
  - action: publish:bitbucket
    id: publish-repository
    name: Publish Repository to Bitbucket
    input:
      repoUrl: 'bitbucket.org?repo=newprojectname&workspace=workspacename&project=projectname'
      description: 'My new project'

By default the project will be created as a private repository. It can be made public using the repoVisibility option.

steps:
  - action: publish:bitbucket
    id: publish-repository
    name: Publish Repository to Bitbucket
    input:
      repoUrl: 'bitbucket.org?repo=newprojectname&workspace=workspacename&project=projectname'
      repoVisibility: 'public'

By default the repository is created with a “master” branch. If you would like to use “main” instead you can us the defaultBranch option.

steps:
  - action: publish:bitbucket
    id: publish-repository
    name: Publish Repository to Bitbucket
    input:
      repoUrl: 'bitbucket.org?repo=newprojectname&workspace=workspacename&project=projectname'
      defaultBranch: 'main'

By default the repository will be populated with the files contained in the workspace directory. If you need to use a subdirectory, you can use the sourcePath option.

steps:
  - action: publish:bitbucket
    id: publish-repository
    name: Publish Repository to Bitbucket
    input:
      repoUrl: 'bitbucket.org?repo=newprojectname&workspace=workspacename&project=projectname'
      sourcePatch: './repoRoot'

Outputs

The publish:bitbucket action produces the following outputs.

Name Description
remoteUrl Url for the newly created repository
repoContentsUrl Url that shows the contents of the repository

catalog:register

This action manually registers a component with the catalog.

You may want to do this if you haven’t configured autodiscovery of components or if you’re using a filename which doesn’t match your autodiscovery pattern.

It has two sets of options. The first allows you to configure the location as a complete url through catalogInfoUrl.

steps:
  - action: catalog:register
    id: register-with-catalog
    name: Manually register with the catalog
    input:
      catalogInfoUrl: https://github.com/RoadieHQ/sample-service/blob/master/catalog-info-1.yaml
      # optional: false # default

The second allows you to configure the repo containing the catalog file through repoContentsUrl and optionally a filepath through catalogInfoPath . You might use this along with the publish:github action.

steps:
  - action: catalog:register
    id: register-with-catalog
    name: Manually register with the catalog
    input:
      repoContentsUrl: ${{ steps["publish-repository"].output.repoContentsUrl }}
      # catalogInfoPath: catalog-info.yaml # default
      # optional: false # default

In both cases you can pass an optional flag which determines if the location can be created before the catalog files exists.

catalog:write

This action creates a catalog-info.yaml file into the workspace directory. It takes an object that will be serialized as YAML into the body of the file.

steps:
  - action: catalog:write
    id: create-catalog-info-file
    name: Create catalog file
    input:
      entity:
        apiVersion: backstage.io/v1alpha1
        kind: Component
        metadata:
          name: test
          annotations: {}
        spec:
          type: service
          lifecycle: production
          owner: default/owner

If you would like to create the catalog file in a custom location you can do that with the filePath option.

steps:
  - action: catalog:write
    id: create-catalog-info-file
    name: Create catalog file
    input:
      filePath: '.backstage/catalog-info.yaml'
      entity:
        apiVersion: backstage.io/v1alpha1
        kind: Component
        metadata:
          name: test
          annotations: {}
        spec:
          type: service
          lifecycle: production
          owner: default/owner

Outputs

The catalog:write action does not have any outputs.

fs:delete

This action deletes items in the workspace. It has one input parameter files that can be provided an array of items to delete.

steps:
  - action: fs:delete
    id: delete-filds
    name: Delete files
    input:
      files:
        - files/deleteme
        - otherfiletodelete

Outputs

The fs:delete action does not have any outputs.

fs:rename

This action allows you to move files within the workspace. The files option takes an array of objects containing from and to options.

steps:
  - action: fs:rename
    id: rename-files
    name: Rename files
    input:
      files:
        - from: copyfrom
          to: copyto
        - from: copyfrom1
          to: copyto1

Outputs

The fs:rename action does not have any outputs.

roadiehq:utils:fs:replace

This action replaces found string in files with content defined in input.

Required params:

  • files: Collection of files and their replacing configuration. See structure of collection item below.
  • files[].file: Path to the file to be modified
  • files[].find: A text to be replaced
  • files[].replaceWith: A text to be used to replace above
steps:
  - id: Replace text in file
    name: Replace
    action: roadiehq:utils:fs:replace
    input:
      files:
        - file: './file.1'
          find: 'i_want_to_replace_this'
          replaceWith: ${{ parameters.templated_text }}

Outputs

The roadiehq:utils:fs:replace action does not have any outputs.

roadiehq:utils:fs:parse

Reads a file from the workspace and optionally parses it.

Params:

Name Description
path* Path to the file to read.
parser Optionally parse the content to an object. (yaml, json, multiyaml)
spec:
  owner: roadie
  type: service
  parameters:
    - title: Path
      properties:
        path:
          title: Path to the file
          type: string
          description: The path you want to get on your backstage instance
        parser:
          type: 'string'
          enum: ['yaml', 'json', 'multiyaml']

  steps:
    - id: roadiehq-utils-fs-parse
      name: backstage request
      action: roadiehq:utils:fs:parse
      input:
        path: ${{ parameters.path }}
        parser: ${{ parameters.parser }}

Outputs

The roadiehq:utils:fs:parse action produces one output.

Name Description
content Content of the file

roadiehq:utils:serialize:yaml

Allows performing serialization on an object

Params:

Name Description
data* Input data to perform serialization on.
replacer Replacer array
options YAML stringify options

options:

Name Description
indent* (default: 2) - indentation width to use (in spaces)
noArrayIndent (default: false) - when true, will not add an indentation level to array elements
skipInvalid (default: false) - do not throw on invalid types (like function in the safe schema) and skip pairs and single values with such types
flowLevel (default: -1) - specifies level of nesting, when to switch from block to flow style for collections. -1 means block style everwhere
sortKeys (default: false) - if true, sort keys when dumping YAML. If a function, use the function to sort the keys
lineWidth (default: 80) - set max line width. Set -1 for unlimited width
noRefs (default: false) - if true, don’t convert duplicate objects into references
noCompatMode (default: false) - if true don’t try to be compatible with older yaml versions. Currently: don’t quote “yes”, “no” and so on, as required for YAML 1.1
condenseFlow (default: false) - if true flow sequences will be condensed, omitting the space between a, b. Eg. ‘[a,b]’, and omitting the space between key: value and quoting the key. Eg. ’{“a”:b}’ Can be useful when using yaml for pretty URL query params as spaces are %-encoded.
quotingType (’ or ”, default: ’) - strings will be quoted using this quoting style. If you specify single quotes, double quotes will still be used for non-printable characters.
forceQuotes (default: false) - if true, all non-key strings will be quoted even if they normally don’t need to.
steps:
  - id: roadiehq-utils-serialize-yaml
    name: serialize yaml
    action: roadiehq:utils:serialize:yaml
    input:
      data: { 'foo': 'bar' }
      options:
        noArrayIndent: true

Outputs

The roadiehq:utils:serialize:yaml action produces one output.

Name Description
serialized Output result from serialization

roadiehq:utils:serialize:json

Allows performing serialization on an object

Params:

Name Description
data* Input data to perform serialization on.
replacer Replacer array
space Space character
steps:
  - id: roadiehq-utils-serialize-json
    name: serialize json
    action: roadiehq:utils:serialize:json
    input:
      data: { 'foo': 'bar' }
      replacer:
        - foo
        - bar
      space: '#'

Outputs

The roadiehq:utils:serialize:json action produces one output.

Name Description
serialized Output result from serialization

roadiehq:utils:jsonata

Allows performing JSONata operations and transformations on input objects and produces the output result as a step output.

Params:

Name Description
data* Input data to be transformed
expression* JSONata expression to perform on the input
steps:
  - id: transform
    name: Transform with jsonata
    action: roadiehq:utils:jsonata
    input:
      data: foo
      expression: <JSONata expression to perform on the input>

Outputs

The roadiehq:utils:jsonata action produces one output.

Name Description
result Output result from JSONata

roadiehq:utils:jsonata:yaml:transform

Allows performing JSONata operations and transformations on a YAML file in the workspace. The result can be read from the result step output.

Params:

Name Description
path* Input path to read yaml file
expression* JSONata expression to perform on the input
options YAML stringify options

options:

Name Description
indent* (default: 2) - indentation width to use (in spaces)
noArrayIndent (default: false) - when true, will not add an indentation level to array elements
skipInvalid (default: false) - do not throw on invalid types (like function in the safe schema) and skip pairs and single values with such types
flowLevel (default: -1) - specifies level of nesting, when to switch from block to flow style for collections. -1 means block style everwhere
sortKeys (default: false) - if true, sort keys when dumping YAML. If a function, use the function to sort the keys
lineWidth (default: 80) - set max line width. Set -1 for unlimited width
noRefs (default: false) - if true, don’t convert duplicate objects into references
noCompatMode (default: false) - if true don’t try to be compatible with older yaml versions. Currently: don’t quote “yes”, “no” and so on, as required for YAML 1.1
condenseFlow (default: false) - if true flow sequences will be condensed, omitting the space between a, b. Eg. ‘[a,b]’, and omitting the space between key: value and quoting the key. Eg. ’{“a”:b}’ Can be useful when using yaml for pretty URL query params as spaces are %-encoded.
quotingType (’ or ”, default: ’) - strings will be quoted using this quoting style. If you specify single quotes, double quotes will still be used for non-printable characters.
forceQuotes (default: false) - if true, all non-key strings will be quoted even if they normally don’t need to.
steps:
  - id: transform
    name: Transform YAML
    action: roadiehq:utils:jsonata:yaml:transform
    input:
      path: a/b/test.txt
      expression: <JSONata expression to perform on the input>
      options:
        noArrayIndent: true

Outputs

The roadiehq:utils:jsonata:yaml:transform action produces one output.

Name Description
result Output result from JSONata yaml transform

roadiehq:utils:jsonata:json:transform

Allows performing JSONata operations and transformations on a JSON file in the workspace. The result can be read from the result step output.

Params:

Name Description
path* Input path to read yaml file
expression* JSONata expression to perform on the input
replacer Replacer array
space Space character
steps:
  - id: transform
    name: Transform JSON
    action: roadiehq:utils:jsonata:json:transform
    input:
      path: a/b/test.txt
      expression: <JSONata expression to perform on the input>
      replacer:
        - foo
        - bar
      space: '#'

Outputs

The roadiehq:utils:jsonata:json:transform action produces one output.

Name Description
result Output result from JSONata json transform

json:merge

Merge new data into an existing JSON file.

Params:

Name Description
path* Path to existing file to append.
content* This will be merged into to the file. Can be either an object or a string.
steps:
  - id: merge
    name: JSON merge
    action: json:merge
    input:
      path: foo
      content: bar

Outputs

The json:merge action produces one output.

Name Description
path Path to the file that got appended to

roadiehq:utils:json:merge

Merge new data into an existing JSON file.

Params:

Name Description
path* Path to existing file to append.
content* This will be merged into to the file. Can be either an object or a string.
steps:
  - id: merge
    name: JSON merge
    action: roadiehq:utils:json:merge
    input:
      path: foo
      content: bar

Outputs

The roadiehq:utils:json:merge action produces one output.

Name Description
path Path to the file that got appended to

roadiehq:utils:merge

Merges data into an existing structured file.

Params:

Name Description
path* Path to existing file to append.
content* This will be merged into to the file. Can be either an object or a string.
options* YAML stringify options (for YAML output only)

options:

Name Description
indent* (default: 2) - indentation width to use (in spaces)
noArrayIndent (default: false) - when true, will not add an indentation level to array elements
skipInvalid (default: false) - do not throw on invalid types (like function in the safe schema) and skip pairs and single values with such types
flowLevel (default: -1) - specifies level of nesting, when to switch from block to flow style for collections. -1 means block style everwhere
sortKeys (default: false) - if true, sort keys when dumping YAML. If a function, use the function to sort the keys
lineWidth (default: 80) - set max line width. Set -1 for unlimited width
noRefs (default: false) - if true, don’t convert duplicate objects into references
noCompatMode (default: false) - if true don’t try to be compatible with older yaml versions. Currently: don’t quote “yes”, “no” and so on, as required for YAML 1.1
condenseFlow (default: false) - if true flow sequences will be condensed, omitting the space between a, b. Eg. ‘[a,b]’, and omitting the space between key: value and quoting the key. Eg. ’{“a”:b}’ Can be useful when using yaml for pretty URL query params as spaces are %-encoded.
quotingType (’ or ”, default: ’) - strings will be quoted using this quoting style. If you specify single quotes, double quotes will still be used for non-printable characters.
forceQuotes (default: false) - if true, all non-key strings will be quoted even if they normally don’t need to.
steps:
  - id: merge
    name: JSON merge
    action: roadiehq:utils:merge
    input:
      path: foo
      content: bar
      options:
        noArrayIndent: true

Outputs

The roadiehq:utils:merge action produces one output.

Name Description
path Path to the file that got appended to

roadiehq:utils:fs:write

Creates a file with the content on the given path

Params:

Name Description
path* Relative path to the file
content* Content of the file
steps:
  - id: create
    name: Create file
    action: roadiehq:utils:fs:write
    input:
      path: foo
      content: bar

Outputs

The roadiehq:utils:fs:write action produces one output.

Name Description
path Path to the newly created file

roadiehq:utils:zip

Zips the content of the path

Params:

Name Description
path* Relative path to the file
outputPath* The name of the result of the zip command
steps:
  - id: zip
    name: Zip the workspace
    action: roadiehq:utils:zip
    input:
      path: foo
      outputPath: bar

Outputs

The roadiehq:utils:zip action produces one output.

Name Description
outputPath Path to the newly created zip

roadiehq:utils:fs:append

Append content to the end of the given file, it will create the file if it does not exist.

Params:

Name Description
path* Path to existing file to append.
content* This will be appended to the file
steps:
  - id: append
    name: Append to file
    action: roadiehq:utils:fs:append
    input:
      path: foo
      content: bar

Outputs

The roadiehq:utils:fs:append action produces one output.

Name Description
path Path to the file that got appended to

fs:append

Appends text to a file within the workspace

Params:

Name Description
file* The path of the file append to. This must be a path within the workspace.
text* The text to append to the file.
steps:
  - id: append
    name: Append to file
    action: fs:append
    input:
      file: ${{ parameters.file }}
      text: ${{ parameters.text }}

Outputs

The fs:append action does not produce outputs.

fs:read

Reads a file from the workspace

Params:

Name Description
path* Path to the file to read.
steps:
  - id: read
    name: Reads a file
    action: fs:read
    input:
      path: ${{ parameters.path }}

Outputs

The fs:read action produces one output.

Name Description
content Content of the file

github:actions:dispatch

The github:actions:dispatch action allows you to trigger the execution of a GitHub action on a repository. The repoUrl option is a repo url for GitHub. The RepoUrlPicker documented above can generate this value. The workflowId can be the workflow id from the GitHub API or you can just use the filename for the workflow file itself. The branchOrTagName indicates which commit to run the workflow against.

This example will run the workflow defined in the “my-workflow-file.yaml” file on the “newreponame” repository on the “main” branch.

steps:
  - action: github:actions:dispatch
    id: trigger-build
    name: Trigger Build
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      workflowId: 'my-workflow-file.yaml'
      branchOrTagName: 'main'

If the workflow takes additional inputs, you can pass these along with the workflowInputs option.

steps:
  - action: github:actions:dispatch
    id: trigger-build
    name: Trigger Build
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      workflowId: 'my-workflow-file.yaml'
      branchOrTagName: 'main'
      workflowInputs:
        parameter1: value1
        parameter2: value2

Outputs

The github:actions:dispatch action does not have any outputs.

github:webhook

You can configure a webhook on an existing repository in GitHub using this action. It takes repoUrl and webhookUrl. The repoUrl option needs to be in a GitHub repo format. The RepoUrlPicker documented above will generate a URL in the correct format.

steps:
  - action: github:webhook
    id: add-webhook
    name: Add Webhook
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      webhookUrl: 'https://webhook-handler-service.abc/handle-webhook'

You can configure a webhook secret using the webhookSecret option. You will likely want to provide this via an output from a previous step.

steps:
  - action: github:webhook
    id: add-webhook
    name: Add Webhook
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      webhookUrl: 'https://webhook-handler-service.abc/handle-webhook'
      webhookSecret: 'mysupersecretwebhooksecret'

You can configure the types of events that trigger the webhook. For a full list of options see here

steps:
  - action: github:webhook
    id: add-webhook
    name: Add Webhook
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      webhookUrl: 'https://webhook-handler-service.abc/handle-webhook'
      events:
        - push
        - pull_request

If you would like the webhook to receive every event, you can set the events to contain ”*“.

steps:
  - action: github:webhook
    id: add-webhook
    name: Add Webhook
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      webhookUrl: 'https://webhook-handler-service.abc/handle-webhook'
      events:
        - '*'

By default the payload of the webhook is form encoded, if you prefer json you can use contentType: json

steps:
  - action: github:webhook
    id: add-webhook
    name: Add Webhook
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      webhookUrl: 'https://webhook-handler-service.abc/handle-webhook'
      contentType: json

You can disable SSL on the webhook request using the insecureSsl option, but it is not advised.

steps:
  - action: github:webhook
    id: add-webhook
    name: Add Webhook
    input:
      repoUrl: 'github.com?repo=newreponame&owner=AcmeInc'
      webhookUrl: 'https://webhook-handler-service.abc/handle-webhook'
      insecureSsl: true

Outputs

The github:webhook action does not have any outputs.

http:backstage:request

This action allows the Scaffolder task to run an HTTP request against the Backstage Backend API and handle the response. It can be useful for extending the scaffolder to call out to third party APIs. You can do this by configuring a proxy and then calling the proxy with this action.

The path should always point to a proxy entry with the following format: /proxy/<proxy-path>/<external-api-path> - i.e.: /proxy/snyk/org/<some-org>/projects or /proxy/circleci/api/projects (NB: the CircleCI proxy path is circleci/api/ but Snyk is just snyk/)

steps:
  - action: http:backstage:request
    id: http-request
    name: Create a thing on the acme service
    input:
      method: GET
      path: '/proxy/snyk/org/<some-org>/project/<some-project-id>'

You can optionally add request params.

steps:
  - action: http:backstage:request
    id: http-request
    name: Create a thing on the acme service
    input:
      method: POST
      path: '/proxy/acme/thing'
      params:
        state: 'bar'

The headers parameter allows setting headers on the request:

steps:
  - action: http:backstage:request
    id: http-request
    name: Create a thing on the acme service
    input:
      method: GET
      path: '/proxy/circleci/api/projects'
    headers:
      Accept: 'application/json'

The body param allows you to set a request body. This is most likely going to be useful for POST requests.

steps:
  - action: http:backstage:request
    id: http-request
    name: Create a thing on the acme service
    input:
      method: POST
      path: '/api/proxy/acme/thing'
      body: 'thingname=abc1'

You can also have the action generate a json formatted body by setting a custom “Content-Type” header to “application/json” and then providing an object to the body param.

steps:
  - action: http:backstage:request
    id: http-request
    name: Create a thing on the acme service
    input:
      method: POST
      path: '/api/proxy/acme/thing'
      headers:
        'Content-Type': 'application/json'
      body:
        thingname: 'foo'

Outputs

The http:backstage:request action has three outputs.

Name Description
code Status code of the http response
headers Dictionary containing all of the response headers and their values
body Body of the response

If there is a content-type header containing application/json the body param will contain the parsed object. Otherwise, it will contain an object with a single param message containing a string representing the body of the response.

debug:log

Use the debug:log action to print some information to the task console.

steps:
  - action: debug:log
    id: debug-log
    name: Log Hello World
    input:
      message: 'Hello, World!'

Outputs

The debug:log action does not have any outputs.

Conditional Steps

You can conditionally execute a scaffolder based on an input parameter.

steps:
  - action: debug:log
    id: debug-log
    if: ${{ parameters.name }}
    name: Log Hello World
    input:
      message: 'Hello, ${{ parameters.name }}!'

Other Actions

You can find all the actions available to your Backstage instance by visiting the following page from within Backstage:

https://<tenant-name>.roadie.so/create/actions

Advanced

Calling an internal API

If you need a scaffolder step to contact a custom authenticated service or any public API for that matter that is not currently supported by a built-in action, you can do that using a combination of the http:backstage:request action and a backstage proxy configuration.

Start by creating a proxy configuration as described in this page

Then you can add a step to call that API using the http:backstage:request action as follows:

steps:
  - action: http:backstage:request
    id: http-request
    name: Create a thing on the acme service
    input:
      method: POST
      path: "/api/proxy/acme/thing"
  - action: debug:log
    id: log-result
    name: Log the result of creating the thing
    input:
      message: "The response code was ${{ steps["http-request"].output.code }}'

Escaping syntax

If you need to pass variable substitution syntax through without it being interpreted you can escape the syntax by wrapping it like so ${{ '${{ parameters.something }}' }}.

Creating re-usable snippets

You can inject in re-usable snippets of yaml into a template using the $yaml operator like so:

templates/debug-step.yaml

- name: Debug log 2
  id: debug_log_2
  action: 'debug:log'
  input:
    message: Second log

logging-template.yaml

apiVersion: scaffolder.backstage.io/v1beta3
kind: Template
metadata:
  name: placeholder-example
  title: Demonstrating the placeholder usage
  description: Shows how to inject in a single re-usable step
spec:
  owner: default/engineering
  type: service
  steps:
    - name: Debug log 1
      id: debug_log_1
      action: 'debug:log'
      input:
        message: First log

      $yaml: https://github.com/yourOrg/some-repo/blob/templates/debug-step.yaml

NB: This can only be done for a single step as the re-usable section must be valid yaml.

Testing

Testing of templates is not well supported in Backstage currently, mostly due to the fact that many scaffolder actions perform side-effects.

A limited set of functionality exists to preview and edit parameters in a sandbox, and dry-run templates (skipping steps that perform mutations).

You can find these features at /create/edit.

It is also possible to test templates by changing the name and namespace of the template to indicate that it is a preview or test version, then adding it to the catalog via /register-existing-component using the version on a published feature branch. This preview template will show up in the list of templates however so it is important to remove the entity after testing to avoid duplication, and also to make sure the title/description indicates that it is a temporary test.

Troubleshooting

Writing templates can be a little cumbersome at times. We have compiled a list of errors that we have seen in the past, that might help you determine the cause of your issue.

Template YAML input forms can also be tested at /tools using a live template preview viewer which speeds up the testing cycle.

preview-template

Resource not accessible by integration

This error is referring to actions that interact GitHub. It means that the Roadie GitHub app is unable to read, create or update the resource/s that are being touched by the Scaffolder step.