Why the description carries the whole thing
Only the name and description are loaded at startup, for every installed skill. The body is read after the agent has already decided to activate the skill. That means a perfect set of instructions behind a vague description is dead weight: it costs context on every session and is never used.
What this cannot tell you
The spec checks are objective and a failure there is a real failure. The heuristics are judgement, not rules: a description can pass all four and still miss the words your team actually uses. Test the trigger by describing the task the way a colleague would, not the way you wrote it.