Lasker's Spanish Inquisition (part five)

LASKER'S opponent in this 1908 Dutch simul, Benjamin Leussen, was a strong master who recorded wins in proper games - not simuls - against Aron Nimzowitsch, Rudolf Spielmann, Max Lange and other celebrated players.
Lasker - Luessen
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.d4 exd4 6.Qxd4 Qxd4 7.Nxd4 c5 8.Ne2 Bd6
Steinitz, 14 years earlier, had preferred 8...Bd7.
9.Nd2 b5!?
This is the first time in this series we have seen someone pursue against Lasker the type of plan popular at club level, namely to push the queenside pawns. Lasker's contemporaries seem, on the whole, to have feared the pawns would become weak by being fixed as targets for White's knights.
10.Nb3 c4!? 11.Nbd4 Bb7 12.f3 Ne7 13.Bf4 0-0-0 14.Bxd6 Rxd6
Stockfish10 and Komodo10 agree this is better than accepting a backward pawn with 14...cxd6. Either way, Black has lost the bishop-pair.
15.c3?!
But Black, with his lead in development, gets something of an initiative after this. More solid was 15.0-0-0.

How does Black grab the initiative (you need to find his next two moves)?

*****
*****
*****
*****
*****
15...c5 16.Nc2 f5
Opening lines to the white king.
17.exf5 Nxf5 18.Rd1 Rd3?!
This looks good at first glance, but after White's reply, Black loses a tempo compared with the direct 18...Rxd1+.
19.Nf4 Rxd1+
19...Re8+ drives White's king to the kingside, which is where it wants to be to support his kingside pawns.
20.Kxd1 Nh4 21.Rg1 g5 22.Ne2
The engines prefer 22.Ne6!?, and if 22...Nxg2!, then 23.Nxc5 Bxf3+ 24.Kd2, when Black has won a pawn but the engines reckon White's active knights give good compensation.
22...h5 23.Ne1 Ng6 24.Rf1 Ne5!?
Stronger, according to the engines, was 24...h4 or 24...Rf8.
25.Nc1?!
White needed to create counterplay. The engines suggest 25.h4 or 25.Ng3.
25...Rf8!?
This keeps an advantage for Black, but even stronger was the obvious 25...Rd8+, when 26.Kc2 loses to Stockfish10's 26...Ng4! 27.fxg4 Be4+. So White has to play 26.Ke2, when the engines continue 26...Ng6 27.g3 Kc7, reckoning Black is well on top (but there does not seem to be an immediate knockout blow).
26.h3 g4 27.hxg4 hxg4 28.Ke2 Re8 29.fxg4?
The king had to get off the e file.
29...Nxg4+!?
Even better, according to the engines, is 29...Nd3 30.Kd2 Nxb2.
30.Kd2 Rd8+ ½–½
This looks like another lucky escape for Lasker - Komodo10 gives Black the upper hand, while Stockfish10, which admittedly tends to be more extreme in its evaluations, reckons Black is winning. Not for the first time in Lasker's simul career, an opponent did not feel confident enough to try for a win in an endgame.