In Scope

• User clicks "Add Step" and the dialog appears showing a list of steps
• Clicking "Use Step" adds a the step to the end of the step list
• User can see the form for the step in the editor
• List of available steps is driven by the API provided by JENKINS-38584
• An extension point for a step editor to plug into
• Implement for at least 2 different steps (so we know what a step editor form looks like)
• User can click on a step after it has been added and edit contents (via its editor, unless it is "unsupported" type see https://issues.jenkins-ci.org/browse/JENKINS-38590)

Out of Scope

• Adding a step which doesn't have an editor (handled in another ticket)

Some more details:

A chunk of steps in the model/json looks like this:

"steps":       [
                {
          "name": "sh",
          "arguments":           {
            "value": "cat /usr/local/apache2/conf/extra/httpd-userdir.conf"
          }
        },
                {
          "name": "sh",
          "arguments":           {            
            "value": "echo \"The answer is 42\""
          }
        }
      ]

(referring to https://github.com/jenkinsci/pipeline-model-definition-plugin/blob/master/src/test/resources/json/agentDocker.json - details may change a bit).

Note that each step object has a name (which can be looked up to see if there is an editor for it). In this case it is the most common sh step - the arguments here is a simple object which has a "value" of the script.

This AST/model is formally defined here: https://github.com/jenkinsci/pipeline-model-definition-plugin/blob/master/src/main/resources/ast-schema.json. This is an example of a "named argument" - there are few ways arguments for steps can be expressed but this is most common (and fairly simple).
Different steps may have different representations: one of namedArgumentList, positionalArgumentList or singleArgument.