wtf is better-thinking?

Better Thinking is a plugin for Claude Code that gives the AI a library of over 130 explicit thinking procedures to use instead of just being told to think harder. The idea behind it is that asking a model to think step by step does not reliably make it more careful, but giving it an actual structured method, like how to properly compare competing hypotheses or how to trace a bug back to its root cause, does. The plugin includes a dispatcher that figures out which of its 135 thinking procedures fits the task at hand, so the user does not need to memorize any skill names. A person just types a single command to start, and it looks at what is being asked, decides whether it is a decision, a diagnosis, research, or something else, and pulls in the right method at a depth that matches how much is riding on the answer. A quick low stakes question gets a fast check, while a costly and hard to reverse decision gets a fuller structured treatment. Each skill in the library encodes a step by step procedure rather than a persona to imitate. The skills are organized into two layers: atomic skills that implement one reasoning move on their own, such as forming hypotheses or steelmanning an argument, and composite skills that chain several atomic skills together into a larger workflow, such as a full scientific method or decision analysis process. This structure means improving one atomic skill benefits every composite workflow that relies on it. The project also ships a set of recipes that map well known frameworks people already use, like Design Thinking or Lean Startup, onto sequences of these skills stage by stage. It installs as a Claude Code plugin directly from GitHub using a marketplace command, and updates take effect the next time a session starts. This is aimed at anyone using Claude Code who wants more rigorous, checklist style reasoning on hard tasks rather than a generic think harder instruction.