I'm not an experienced French defence player but I might give this a shot anyway.
3... c5 might be a book move in a sense that people have tried to play it here because it's a logical try to undermine white's centre but it does not have good reputation in this particular position. I'm not super familiar with the details but probably the main problem is that white gets too much immediate pressure against the black central pawns. Against 3. Nd2 the move 3... c5 has a much better reputation and is a popular reply. You could try to look from database how white plays after 3. Nc3 c5?!.
I guess 5... Bd7? is bad because d5 is left undefended. 6. exd5 exd5 7. Nxd5 would simply win a pawn.
16... Re8?! 17. Bxf6+ Kg6 18. Qxe8+ wins more material for white.
So I’ve decided to give the French defense the good 'ol college try. This is one of my first games using it, the engine called it the French Defense Marshall Variation. I was on the ropes early but pulled out a win via my opponents blundered queen. E5 was my lone blunder, I saw a pawn fork and just went for it. Just a few questions:
Why was 3.c5 an inaccuracy?
I thought c5 was a book move.
Why was 7.bd7 a mistake?
Why was 16.re8 an inaccuracy?
Would love to get thoughts on this game from experienced French Defense players. I still have trouble deciphering why the engine likes and dislikes certain moves. Feel free to roast my game or add any constructive criticism.