Although the scenario being described is in a different location on the field, this question appears to be the similar to this related Q&A:
Since Entanglement is not occurring, G12d should only be used as the general guideline that Cubes are an extension of their Robot. So, picture this interaction as if the Cubes were just square metal mechanisms resting on a Robot.
- Bumping into opponents is not illegal in itself. However, there is a threshold where bumping becomes reckless ramming, and could be considered a violation of G12, S1, or G1.
- Dislodging parts of your opponent's robot (without Entangling them) is not illegal in itself, because part (c) puts some responsibility on the opponent to design their robot such that it is not easily damaged by minor contact. However, there is a threshold where incidental damage becomes intentional/egregious destruction, and could be considered a violation of G12, S1, or G1.
With regard to the following portion of your question:
The cubes are not scored yet but are being descored with this action.
Match Affecting verbiage should only come into effect when a Head Referee has called a rules violation, and the rule in question requires them to determine whether the Robot should receive a warning or a Disqualification via Match Affecting verbiage.
The scenario described in this question is not, in itself, a violation, provided no other rules are being violated. If the Head Referee does determine that G12 has been violated in an egregious manner, then Match Affecting verbiage wouldn't apply anyway, since G12 is one of the only rules which does not hinge upon "Match Affecting".
The interpretation of this rule and the amount of defense tolerated is different everywhere we go!
Ultimately, we must again refer to the previously linked Q&A:
A volunteer Head Referees' primary responsibility should be to interpret the rules as written and as the context of their event necessitates. If a game is played more defensively at a different event, then it would stand to reason that the judgment calls would be handled accordingly in the context of that event.
We would add that the dynamic sport of competitive robotics inherently includes "gray area" situations that cannot rely on black-and-white rules, as do many other sports, and must therefore rely upon the in-context and on-the-spot judgment of human referees in these cases. Any time an on-the-spot judgment is needed by a volunteer Head Referee, there is a possibility that two different volunteers may interpret the situation differently.
Our best advice to Teams who are struggling with this variance would be to compete in a manner which minimizes the risk of relying on a judgment call (especially a G12 one).