Consider a stationary Poisson process $η$ in the $d$-dimensional Euclidean or hyperbolic space and construct a random graph with vertex set $η$ as follows. First, each point $x\inη$ is connected by an edge to its nearest neighbour, then to its second nearest neighbour and so on, until $x$ is contained in the convex hull of the points already connected to $x$. The resulting random graph is the so-called nearest neighbour embracing graph. The main result of this paper is a quantitative description of the Gaussian fluctuations of geometric functionals associated with the nearest neighbour embracing graph. More precisely, the total edge length, more general length-power functionals and the number of vertices with given outdegree are considered.