It's worth to note that even for a conservative holonomic system the kinetic energy is a function of both the co-ordinates and generalised velocities, i.e. the kinetic energy is a homogenous quadratic form with position dependent coefficients,$$T(\boldsymbol{q}, \dot{\boldsymbol{q}}) = a_{jk}(\boldsymbol{q})\dot{q}^j \dot{q}^k \quad \mathrm{where} \quad a_{jk}(\boldsymbol{q}) = \frac{1}{2}\sum_a m_a \left( \frac{\partial \boldsymbol{x}_a}{\partial q^j} \cdot \frac{\partial \boldsymbol{x}_a}{\partial q^k}\right)$$in which case the Lagrangian has the form ##L(\boldsymbol{q}, \dot{\boldsymbol{q}}) = T(\boldsymbol{q}, \dot{\boldsymbol{q}}) - V(\boldsymbol{q})##. Further, Lagrange's equation for any holonomic system ##\frac{\mathrm{d}}{\mathrm{d}t} \left( \partial T / \partial \dot{q}^j \right) - \partial T / \partial q^j = Q_j## is formulated only in terms of ##T(\boldsymbol{q}, \dot{\boldsymbol{q}})##.
For more complex holonomic systems with moving constraints [##\boldsymbol{x}_a = \boldsymbol{x}_a(\boldsymbol{q}, t)##], the kinetic energy may also contain a functional time dependence ##T(\boldsymbol{q}, \dot{\boldsymbol{q}}, t) = a_{jk}(\boldsymbol{q}, t) \dot{q}^j \dot{q}^k + b_j(\boldsymbol{q}, t) \dot{q}^j + c(\boldsymbol{q}, t)##. Also, with moving constraints the potential energy now depends on time, ##V = V(\boldsymbol{q}, t)##.
[Edit: and also, as mentioned above, it's possible to have even weirder systems where the potential now also depends on ##\dot{\boldsymbol{q}}##, in which case ##L(\boldsymbol{q}, \dot{\boldsymbol{q}}, t) = T(\boldsymbol{q}, \dot{\boldsymbol{q}}, t) - U(\boldsymbol{q}, \dot{\boldsymbol{q}}, t)##!]