Several people I’ve talked to have tried this, with no success. It’s frustrating, because as you say, the “numberOfChildren” key is a tease, seeming to promise that the “meat” of the action might be fetched and examined. There’s always saving the action to disk, and understanding the resulting file by reverse-engineering its format! If anyone knows a better way here, I’d really love to hear about it.
Compile listner plugin example.
Place it in a photoshop Directory.
Start Photoshop and runn Command that you are interested about. In the listner.log (found usualy on an root of the main disk) you will find function generated by listner that calls Command filling the cal with properties that command needs.
It is then easy ro turn listner code u coresponding get descriptor/properties commands.
Hope this helps
First, the listener output is not identical to the content of an action. But second, that doesn’t help for pulling the content out of an action at “runtime” without running it explicitly. It seems like this should be possible, given the “numberOfChildren” key in the “command” descriptor, but so far seems not to be. So as mentioned, the alternative is to save the action to disk, and then parse the resulting file.