The pluginId in that cluster profile fragment refers to a completely different ec2 elastic agent plugin so not sure what you're trying to show us there 😅
Personally, I don't think it's a good idea to be directly configuring cruise-config.xml especially for plugin-oriented config, since it necessarily cannot be schema validated. Elastic agent config is typically complex with validation rules implemented in the plugin. If automation is the goal, using the APIs for automation is likely a better experience than cruise-config.xml hacking.
If you really want to control cruise config directly, start with a working UI-driven configuration and work backwards to the required cruise-config XML fragment.
Otherwise you'll need to look at the logs and there's no guarantee as to how friendly the error messages will be, as you are shortcutting all of the UI-based validation assistance with that approach.