AI Toolkit

Make AI try to break its own conclusion before it hands you one

A Claude Code skill that forces grounded research and a cheapest-falsifying test before the AI commits to any consequential claim. Instead of confidently telling you what you want to hear, Claude has to state what would prove its claim false, go get real evidence, design the cheapest test that could kill the claim, and report honestly when the data proves it wrong.

Turn a rough task into one /goal line Claude can actually prove

A Claude Code skill that converts a vague task into a single copy-paste /goal command with a measurable end state, transcript-visible proof, explicit constraints, and a turn cap. Claude Code's built-in /goal loop keeps the AI working until a condition is true — but the condition is judged by a small model that can only see what's printed in the chat. A goal like “until the code is clean” never terminates. This skill writes goals that carry their own proof.