I could not find it, but I know this was discussed before.
I don't like the idea for various reasons. I think the cpu you can use should have a cap... and I think it would not be good for the market.
If considered it should have a high cost, but not in energy or credits but in cpu. ( So if you sell 500 cpu it would cost the seller 1000. )
if you want code only run exactly once you could save a flag in Memory for example.
could you please push it to npm ? 
maybe you should look into typescript .. I found it much easier to get started with then javascript .. using classes, inheritance and stuff. I know all that is also possible with pure javascript but typescript helps you to prevent making many stupid errors ,)
https://github.com/screepers/screeps-typescript-starter
also look for help in slack
@thynosaur and that would be an object.. since there can be several types of resources in a store.
ah .. I misunderstood the images... yes .. I had a similar problem once .)
I assume the tower not shooting on his creeps is a bug in your code... you should always have as many towers as possible for the level of your room.
you don't need attack parts to destroy structures, because you can dismantle them with work parts. you only need attack parts to attack other creeps.
And a wall of 1000 is really nothing .)
I guess with standard moveTo it will probably work because it will redo the path every few ticks .. so it's ok to just get the right direction for the next 5 steps.
I would also prefer store .. it's short and I think it fits.
@artch would it be possible from the users script to modify the storage object of a creep and reset the cache? That would be nice because I currently already patch the creep object for example on successful withdraws, pickups and stuff like that so decisions made after that are based of the new values ( I know those can still fail when multiple creeps for example withdraw, so that should not be automatic maybe ) but I use it only for the decision to move to the creeps next target for example in the same tick.
If this is a breaking change ( and there are already other mentions in the docs about deprecated stuff like the transfer method on structures, the old spawn function,... ) maybe it's time to create a second selectable runtime? I don't know how much work that would be but perhaps it would also allow us to get one with lodash 4 already included ? 