We can do that now. It's fairly easy for anyone to edit viruses or bacteria using non-crispr methods to make something more virulent.

The problem is there is an evolutionary trade-off. To be more virulent means dedicating resources to it, to the detriment of other kinds of fitness. The result is your virulent pathogen cannot spread, defeating the point of a weapon.

We are many decades away from achieving systemic understanding necessary to improve an organisms fitness using molecular tools, and also account for compensatory changes to fitness.

You can however use selective breeding to do this, but that's millenia old technology.

> I define "long way" as 20 years.

We don't even understand most of the existing proteins in databases yet, and it is going to take decades to meticulously characterize these sequences.

Worse, nobody has come close to a useful denovo protein sequence yet, let alone a genomes worth of proteins and their interactions. I feel safe in saying we are at least 50 years off from these goals, probably more.