Which is the best description of how Prim's algorithm finds a minimal spanning tree?

It's an iterative process where in each iteration a cheapest edge is added as long as it forms a connected graph with previously added edges and doesn't produce a cycle. It terminates when any added edge necessarily produces a cycle.

It's an iterative process where in each iteration a cheapest edge is added as long as it doesn't produce a cycle. It terminates when any added edge necessarily produces a cycle. Until the last edge, there is no requirement that an added edge forms a connected graph with the previously added edges.

It's an iterative process where in each iteration a costliest edge is added as long as it doesn't produce a cycle. It terminates when any added edge necessarily produces a cycle. Until the last edge, there is no requirement that an added edge forms a connected graph with the previously added edges.

It's an iterative process where in each iteration a costliest edge is deleted as long as it the resulting graph is connected. It terminates when any edge deletion would necessarily result in a graph that is not connected.

It's an iterative process where in each iteration a cheapest edge is deleted as long as the resulting graph is connected. It terminates when any edge deletion would necessarily result in a graph that is not connected.