Expected behavior:
Place a powered assembler and an inserter facing into the assembler. Place an item (in this case a wooden chest) into the inserter's hand. The inserter's hand swings, and at the end of the swing, hovers over the assembler with its item, as shown below:

Unexpected behavior:
Place a powered assembler and an inserter facing into the assembler. Connect the assembler to a power pole with a wire, then place an item into the inserter's hand. This time the inserter's hand swings, and at the end of the swing, the item is placed into an output slot of the assembler:

Now disconnect that assembler from all circuit networks, and again place an item into the inserter's hand. The item again is placed into an output slot despite the assembler being returned to the state in the first image. The behavior of the assembler can only be restored by breaking and replacing it. In my opinion, this should qualify this behavior as a bug.
In the linked bug report, Muche argues the inserters
I believe the point is this should be enough to prevent this manual storage method from being automated. However, this is not true, as shown in my previous post (viewtopic.php?f=8&t=125185) and demonstration video (https://www.youtube.com/watch?v=sEFFIDZIVO0). Clearly automated infinite storage density is not intended to be possible. The demonstrated design is slow, but it could be made much quicker with slightly more effort and complexity.will not start putting random things into empty machines after they have empty hands, know the machine is there, and does not want those random things.