A few remarks to the game presented by MNb
10...c6 is actually a sideline. The main line continues with Nbc6.
10....Nbc6 11.Rae1,Qf5 12.Nd5,Kd8 13.Qe2,Qe6 14.Qf3
is the Lean-attack.
Old theory books now state that Black should repeat moves with 14...Qf5. But actually 14....Qg6! promises Black good winning chances, as pointed out by S.Buecker/H.Smout digging up old analysis by Snosno-Borovski (1911).

That 18....Qg6 was the first new move in Mnb's game might be clarified a bit.
Both 18....Qxf4 and 18...Nxe7 had been analysed by Schlechter in the Bilguer as early as 1916.
Actually 18...Qxf4 leads to a forced draw as well after
19.Nxf4,Nxe7 20.Qg5+,Kh8 21.Nh5,f5 22.Qe3! (Blankenberg-Hannison,2001) when a perpetual is set up with 23.Qd4,Rf7 24.Qd8+ or 22...Rf7 23.Qe8+.

There is also an interesting side variation with
13.Bxd5,cxd5 14.Qh5,Qd6 15.Nb5! unclear (Samisch-Gunther, 1926). Black might have entered this in the game by replying 14.Bxd5,cxd5 (iso 14...0-0). But this also would not have given black any winning chances imo.

Just two things to be aware of:

• There's already quite a bit of Muzio analysis in the threads on this site, you may want to check those out before seeing whether you're re-inventing the wheel. You can check them out by typing "C37" or Muzio in the search box.
• The Muzio Gambit doesn't lend itself particularly well to engine analysis. There are quite a few sac lines that the engines just completely misevaluate even in these days. (Perhaps Fritz 13+ will prove this statement wrong, but I doubt it.)