How are the canards designed to stall first?
Canards are normally built with a very high aspect ratio: this gives a quite sharp stall and at a quite low AoA. But that's not the whole story. Canard design is a sensitive (and also quite interesting) tradeoff among contrasting characteristics.
Why is canard configuration so sensitive? Let's suppose that, for whatever reason, the nose of the aircraft goes up. In order for the aircraft to be stable, it must respond pushing the nose back down.
When the nose goes up, the AoA of both the canard and the wing increase and their lifts increase accordingly. A bigger lift generates a bigger downwash. Now, the downwash of the wing doesn't bother us that much since behind the wing there's basically nothing important in a canard configuration. But the downwash of the canard does matter to us because it impinges on the wing thus reducing the local AoA of the wing.
So, a nose up movement of the aircraft translates in an increment of the canard's lift plus an increment of the wing's lift although smaller. But bigger lift in front means nose up tendency. Therefore, if the nose of the aircraft goes up, a canard configuration pushes the nose even more up: the canard configuration is fundamentally unstable.
How can this instability be cured? There are basically 3 things which can be done and each one has pros and cons.
The CG is shifted as far as possible from the wing's aerodynamic center (i.e. as forward as possible); this can be achieved for example sweeping the wing back: its weight and its aerodynamic center shift backward and this has the same effect as shifting the total CG forward toward the canard. But! a more forward CG also implies that a big part of the weight must be now lifted by the canard which therefore has to generate more lift and more downwash, exacerbating at the end the just explained instability.
The canard is built in such a way as to generate always less lift than the wing, also taking into account that its downwash blows on the wing reducing the wing's AoA and lift. This can be achieved for example if: a) the canard has a smaller surface than the wing (1:4 for example for the Rutan VariEze); this normally also implies that the canard's span is smaller than the wing's one and its downwash now impinges on a smaller portion of the wing; b) the canard is built with a very high aspect ratio; this gives a well defined stall and at a lower AoA; c) an airfoil is chosen possessing a quite flat $C_L \text{vs } \alpha$ curve around the stall:

- This plot from this NASA report shows the lift coefficient of the canard of the Rutan VariEze (circles) in respect to a standard NACA 0012 airfoil (squares): as visible, the stall region is well defined for the VariEze while for the NACA 0012 there's a region after the pick where the lift goes down and then up again; furthermore "canard stall occurred at about $\alpha$ = 13°, whereas wing stall occurred at about 21°".
- We take to the extreme the previous solution and we simply download completely the canard; it now generates lift only to control the aircraft and only in a relative small amount and its shape can be now more or less freely chosen; the instability problem due to the downwash is gone but now there's another problem: if the canard does not lift anymore, then the whole weight is supported by the wing alone i.e. the CG is now very close to the aerodynamic center of the wing; unless the wing is designed self-stable, this condition is not only unstable (again) but this instability is normally quite fast as well and the only way to counteract it is via an automatic stability system which is not really an option on small general aviation airplanes.
If you see canard aircraft with a relative big canard surface, think about the point 1.
If you see a Rutan's aircraft, think about the point 2.
And if you see an Eurofighter, think about point 3.