the body-fixed frame
The body-fixed frame is a set of coordinate axes painted onto the rigid body itself, so they rotate and tumble along with it. If you glued a tiny set of x-y-z arrows to a spinning top, those arrows are the body frame. From inside this frame the body never moves -- every atom sits at a fixed coordinate -- while the outside world appears to spin.
The whole point is that in the body frame the mass distribution is constant, so the inertia tensor is a fixed matrix of numbers rather than a time-varying one. Almost always the frame is anchored at the center of mass and aligned with the principal axes, making the inertia tensor diagonal. The catch is that the body frame is non-inertial (it rotates with angular velocity omega), so time derivatives must be converted with the rule (dA/dt)_space = (dA/dt)_body + omega x A for any vector A. That single rule is what turns Newton's rotational law into Euler's equations.
You use the body frame whenever you track a real spinning object -- a satellite, a gyroscope, a molecule. The trade-off is honest bookkeeping: the frame makes the inertia tensor simple but the frame itself is accelerating, so fictitious (centrifugal and Coriolis) effects enter through the omega x A term. To recover the orientation seen by a lab observer you compose the body frame with the space frame using the Euler angles.
The International Space Station carries body-frame gyroscopes and computes its attitude in a frame fixed to the station's structure. Its guidance software constantly converts between that body frame and an inertial frame tied to the stars to know which way the solar panels face.
In the body frame the inertia tensor is constant; the price is a rotating, non-inertial frame.
The body frame and the inertial (space) frame share the same origin only if that origin is the center of mass or a fixed pivot; otherwise you must also track the translation. Confusing quantities measured in the two frames is the single most common error in rigid-body problems.