@MikeMiller $M$ be a smooth manifold, $x_0$ a point on $M$ and $\vec{v}$ a tangent vector on $M$ at $x_0$. Given any smooth function $f$ on $M$, taking directional derivative of $f$ along $\vec{v}$ at $x_0$ spits out a real number. I'm using the word smooth very informally, as I don't know the definition yet, and the only context in which I know what directional derivative of a smooth function over a manifold means, is when the manifold is $\Bbb R^n$.
However, speaking out of intuition, I can bet this map $C(M) \to \Bbb R$ is linear and satisfies the Leibniz rule, as it does so for $M = \Bbb R^n$ too (I expect this to hold as smooth manifolds are by definition locally diffeomorphic to $\Bbb R^n$). Analogizing this, pick an affine variety $V$. $x \in V$ be a point in the variety. Define a tangent vector to be a linear function $k[V]_x \to k$ from the stalk at $x$ satisfying the Leibniz rule.
Collection of these tangent "vectors" gives you the tangent space. But I am most unhappy with this definition, as it doesn't reveal any good geometric/sheafy information about the variety around $x$ - at least not in this formulation.