I have a feeling we might need to come back to this pattern again in the future; as I imagine we'd want to use the same rig for classnames as well as variables, and it wouldn't be possible to use this current approach for that

i.e.

<<~EOF
class #{rig['class']} {
  public function __wake() {
    $#{rig['payload']} = '...';
  }
}
EOF

Also in some scenarios we might want to have the identifier without the $ prefix, such as when accessing params in PHP - such as $_GET['#{rig['injection_param']}']; etc

Well if the $ acts as a namespace, then maybe the language should have been named php-vars or something so we could add php-class because you could have a class named Foobar and a variable named $Foobar with separate values right? So it doesn't necessarily make sense to track them together here. For code readability it'd make sense to not reuse the identifiers, but obfuscation is part of the point here, so all the better if names are reused but still valid.