The search box of rules is quite smart. It inspects rule names, descriptions, tags and languages (even more, I didn't check). That means that you can type "misra" (or "MISRA"), and all rules tagged "misra" will be returned. It can be considered as a shortcut over usage of tags facet (see attached screenshot)
This search by tag is implemented with an edge n-gram analyzer. It allows to search by any prefix of analyzed text. In our case it allows to type "misr", without the ending "a", and it will (almost) return the rules tagged with "misra" (see on NEXT). Same for "mis".
This an unwanted and unused advanced feature. Using a simple "word" analyzer would make search more simple. The edge n-gram analyzer should be replaced with a word analyzer. In the case of tag "brain-overload", that means that all of "brain", "overload" and "brain-overload" would return results (dash is considered as a word separator). "brai", "brain-over", "verload" or "verlo" return nothing.