Q Thinking Chain

Three named reasoning techniques -- reframe past the safe answer, destroy an idea, track competing hypotheses.

Door kju4q · kju4q/q-claude-skills

Getest · Werkt ★ 8.8/10

Q Thinking Chain — Three named reasoning techniques -- reframe past the safe answer, destroy an idea, track competing hypotheses.

Wat het doet

A structured-thinking skill bundling three techniques: a three-layer 'reasoning layer' that forces a committed answer past the safe default, an 'adversarial room' that isolates the single fatal assumption behind an idea, and a 'belief updater' that tracks three competing hypotheses as evidence arrives. Triggers on requests for a genuine opinion, a stress-test or critique of a plan, or decision support under uncertainty.

Testrapport

Ran the adversarial-room technique live on a real pitch (an AI Instagram-caption SaaS for photographers): it isolated one falsifiable fatal assumption ('photographers find captions painful enough to pay for') and named a concrete competitor response, a sharper and more mechanistically faithful result than a hedged pros/cons baseline. The parent repo's own README 'What's inside' catalog never mentions this skill (listed internally as 'Claude Chain') at all, so a user browsing the collection would not discover it exists.

Getest op: 2026-07-15 · Claude Code 2.x (agent harness)

Installatie

git clone https://github.com/kju4q/q-claude-skills.git
cd q-claude-skills
mkdir -p ~/.claude/skills
cp -r skills/claude-chain ~/.claude/skills/q-thinking-chain  # repo folder is named claude-chain; frontmatter name is q-thinking-chain, rename on copy

Commando's en voorbeeldprompts

  • /q-thinking-chainThree named reasoning techniques -- reframe past the safe answer, destroy an idea, track competing hypotheses.

Skills reageren op gewone verzoeken — geen commando's om te onthouden. Na installatie activeren prompts zoals deze de skill (in het Engels):

  • stress test my startup pitch and find the one fatal assumption
  • give me your real opinion here, not the safe diplomatic answer
  • track competing explanations for this bug as new evidence comes in