Misleading doc about the think toolset

[dgageot]

The documentation of the think toolset (https://docker.slack.com/archives/C07TNPD5SG5/p1773987593802989)
says this:

Include the think tool in every agent. It adds minimal overhead while significantly improving reasoning quality on complex tasks.

This is wrong. In a nutshell, this toolset is useful for small llms that don't have reasoning capabilities.
It is the poor man's thinking for non thinking models.