What part of "all syntax should be validated" don't you understand? By putting that syntax in comments, syntax errors will not trigger any output from the tools because they remain valid free-form comments.
And they're not "Go" syntax. The syntax in the comments if for Go tools. A better approach may be to provide a more extensive syntax checking tool that parses comments and checks for syntax errors on a per-tool basis, but as I understand it it would seem that the implementation as it is now is aimed at the auxiliary Go tools, standardizing the syntax for the tools, thus allocating potential for future tools to expand without extensive maintenance on the core language itself. The tools are not the core language, but there should probably be some form of syntax checkers for the tools, but perhaps encapsulated within the tools. The comments are still comments to the actual Go code, as they are not pertinent to the language itself, but rather an outside entity such as a tool.
It is also entirely possible that I am missing the point entirely and do not understand.
It's irrelevant which tool does the checking. The problem is that you don't know whether "// go:generate foo", "//ho:generate bar" or "//go;generate baz" are errors or simply comments.
In order to detect most of that you'll have to redefine the comment syntax for the core language so it's no longer free-form but anything that looks like a comment-directive now triggers a warning (or most likely an error seeing the Go philosophy).
