The power (work per unit time) of a force $\vec F$ on a particle with velocity $v$ is given by:
$$ P = \vec F \cdot \vec v.$$
If this number is positive, then the force tends to speed the particle up. If it is negative, it tends to slow the particle down. You can see this in the Work-Energy theorem, the differential form of which consists of applying Newton's law to this as if it's the only force acting on the particle:
$$ P = m {d\vec v \over dt} \cdot \vec v = m~~\frac{1}{2}~{d \over dt}\left(\vec v \cdot \vec v\right)$$We see that $\vec v\cdot\vec v$ is the speed squared and has no dependencies on anything other than the speed, moreover this is increasing with time when the power is positive, or decreasing when it is negative, or zero when it is zero.
Now, the thing about that dot product, $\vec F \cdot \vec v$, is that it is zero whenever those two vectors are perpendicular. So a force has to have some component in the direction of your velocity.
Magnetic forces do not have this: their Lorentz force is $\vec F = q \vec v\times\vec B$, always perpendicular to velocity. Magnetic forces cannot do work directly. To calculate, say, two magnets attracting each other, you always need to see what sorts of electric fields are getting induced, or at least calculate the energy density in the fields before and after.