GitHubOpen source

Who's doing the thinking?

Paste an AI conversation. See who was doing the thinking.

We don't store your data. Classification uses Anthropic Claude, which retains API inputs for 30 days per their privacy policy. They do not train on your data.

dontkillmybrain measures

Steering

Whether you set the direction, defined constraints, or reframed the problem in your own terms.

Friction

Whether you pushed back on AI output, questioned assumptions, or rejected suggestions.

Contribution

Whether you added original ideas, examples, or context that shaped the conversation.

Evaluation

Whether you critically assessed the AI's output and made a judgment about its quality or correctness.

Sample report

Claude12 human messages

Debugging a React state management issue

You started strong — you described the bug precisely and proposed a hypothesis about stale closures. But midway through, when Claude suggested refactoring to useReducer, you accepted without evaluating whether it actually addressed your root cause...

“Next time you hit a wall, try explaining why you disagree before asking for alternatives.”