[vibrio]

You could get a 'super T cell' that behaves just as you want, or as is empirically observed, mutations would more commonly lead to leukemia. There is a phenomenon call somatic hypermuatation, that as T-cell respond to antigens, they 'get sloppy' in coding their DNA to optimize the response. It' has also been linked to blood cancers, so it is a risky approach, like many aspects of immunity.

[adrianN]

Of course almost all mutations won't be beneficial (this is also true of germline mutations). But "can only be harmful" seems to imply that beneficial mutations are impossible.

[vibrio]

agreed. sort of my point: some changes can be beneficial, but the process is very risky and the cell goes through great measures to minimize any sequence changes. some viruses are sloppy on purpose, as they have tens or hundreds of thousands of offspring, so trial and error is acceptable.