Making the input into a string of triangle-sides is easy with a two-arm 11I mechanism, and from there it's straightforward to fold it into a triangle in 10 or so I. I did not expect to find a way to do so in just 5I.
- reuse the input arm to bond the lead onto the quicksilver
- 4I for an arm that moves and pivots the partial product in position for the bi-arm to grab
- reuse the bi-arm with +1I to pivot and rotate it into the correct position
- somehow the 4I arm is already positioned exactly right to output the product without crashing
I wouldn't be too surprised if there is a better initial assembly than the 11I base used here. Assembling a corner instead of a side seems like it has potential, but best I can do is a 12I assembly with no clear way to combine in 3I.