1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. exd5 exd5 5. Qf3 Qe7+

6. Nge2  (you have this as move 7; it should be move 6 ) 

6...   Nc6  ( this is the best reply )

This is what both Williams and Watson recommend now.  Black gets a very nice game here.

In the Winawer you have a line which goes like 1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.exd5 exd5 5.Qf3 (Larsen) and now Qe7+!. This seems to be good for black, disrupting white's development. Both Dvoretsky in 1995 (Secrets of opening preparation) and recently John Watson (Dangerous weapons: the French) wrote about it, although Watson played the devil's lawyer  (don't know if this is a regular expression in English, it is in Dutch) and avocated the system for white. Simon Williams and, again John Watson, defended the black side in Attacking chess: the French and Play the French, 4th edition. 

Now, in the variation 1.e4 e6 2.d4 d5 3.Nc3 Nf6 4.Bg5 Bb4 5.exd5, nearly everyone playes Qxd5, which is a good move of course. But why not 5...exd5? Because after 6.Qf3! white threatens Bxf6, crippling the black structure. Recently Aagaard and Ntirlis also embraced 5...Qxd5 in their, very good, book about the French. And indeed, doesn't black has to play 6...Be7, losing a tempo, and can't white simply continue with 7.Bd3/Nge2/0-0-0 and/or (7).h3 ?

Well, maybe there is another disruptive move here after 6.Qf3, 6...Qe7+!?. A short analysis without the engine convinced me it seems playable:

A. 7.Be3?! Ne4
B. 7.Be2 Bxc3+! 8.Qxc3 Ne4! (the point) 9.Bxe7 Nxc3 10.bxc3 Kxe7 or 10.Bb4 Nxe2 11.Nxe2= (not very fun, I admit). Or 8.bxc3 Nbd7 (8...Be6) and h6/g5. Maybe 7...Nc6 8.0-0-0 Be6 with a complex position?

C. 7.Ne2 just seems to block the B on f1. 7...Be6 8.Bxf6 gxf6 9.a3 Ba5 of Bd6, complex it seems.

Of course there are many other tempting possibilities in this variation, this is just a start. Don't know if it is a novelty.