> I suggest the Salamander move as a Q3 that must approach a piece whenever it moves.

That's another interesting possibility.

I think the move I settled on is better for what I was trying to do, which was to keep the Salamander as a good defensive piece while reducing its ability to promote. I assume your formulation lets the piece move when it lands next to a piece of either color along its line of movement. This means there's just one direction in which the enabling piece (should we call it a 'screen', by analogy with hopping pieces, even though it's in the wrong place?) can be located, as opposed to 7 in my version, so your Salamander would be less powerful overall. If your screen could be an opposing piece as well as a friendly piece, that means (other things being equal) that your Salamander would be able to make an approach-constrained move about 2/7 as often as mine, and might be less inhibited about moving into enemy territory, because the opponent would typically have a number of pieces that could not capture the adjacent Salamander.

I might use your rule elsewhere, though. I have long thought that the Jupiter Hooklet and Hawklet might be overpowered for that game (I can't really tell until I have some software capable of playing the game). One way to throttle them down would be to make the second parts of their moves conditional on an approach. In their case, because their promotions are not as strong as a Genie, there would be less reason to discourage them from moving toward the promotion zone.

The fact that you can so readily come up with a different, but still easily understood, rule of movement for a piece worth less than a Rook should cheer the folks at the ShortRange project. There is still a very large space of interesting pieces for them to explore.