If you are going to PWM the motor, the capacitor will cause the current to be very large at the beginning of the pwm period. This is undesirable. So, if you keep the wires short, I’d say: try it without a capacitor. If you run into trouble, try the capacitor.
But: you are dealing with effects where connection wires are no longer electrical connections. The capacitor needs to be close to the motor, and the wires from the controller to the motor help keep the peak current low enough not to blow up your motor controller.
Advice for a typical value of the capacitor would vary a lot too. My advice would go against that of david and say: start out with 1nF. When you’re ordering components, just get 10pf, 100pf, 1nF, 10nF, 100nF and 1uF. Get at least ten of each, so that you can experiment with different values.
The reason I’d start out with a much lower value is that in theory (if the wires WERE ideal electrical connections) and if the capacitor were ideal, you’ll exceed the maximum current of the motor driver for a short period. Guaranteed. Luckily both the wire and the capacitor won’t be ideal… (which means you have to put the capacitor away from the motor driver, near the motor!)
On the other hand, you can do some mathematical calculations for the capacitor. If you are PWM-ing at 10kHz, you’ll have to charge and discharge the capacitor each cycle. the energy in a capacitor is .5 C.U.U, At 6V that becomes 18C. With a .1 uF capacitor, that would become 1.8uJ. at 20,000 switching moments per second (once on, once off for each period), that comes to 36mW. All this goes into your motor driver. This 36mW is probably not a problem. I was honestly expecting a bit more…
But as you see, it depends on the switching frequency. If you chose 20 or 30kHz, the number goes up.