Drew's comment points in the general direction, and your response is also apt, so I'll use that context for the following example.
In code/IT, as well as other technical pursuits, it can be the case that the people who develop or sell some item (let's call them ABC) also service (or support) that item. For various reasons, ABC may not want end users (also endusers in US) to actually try to service the item themselves. (Reasons may include
- ABC wants to be paid to service the item.
- ABC fears getting sued by end users who injure themselves attempting to service the item.
and may be multiply applicable.) If so, ABC will not want information that could assist (or be useful for) some enduser to reach them. More generally, when person P says that something is "too useful," typically
- that something is information
- P means that the information would actually be useful to some person/group Q to perform some activity A
- P means that some person/group R does not anyone else (at least, not Q) to perform A, and therefore does not want that information to reach Q
- P might more exactly/tediously say that "R fears that information would allow Q to do A," but as noted above, persons who use "too useful" typically intend to be droll.