Slow Thinking: Let AI Be Your Practice Partner, Not Your Substitute

Slow Thinking is a Claude Code skill that transforms AI from an "answer provider" to a "thinking training opponent" through a three-stage training process and 10 thinking cards. It helps users exercise their independent thinking skills in 5 minutes a day, combating the problem of thinking degradation caused by "outsourcing thinking" in the AI era.

The rapid development of large language models brings a paradox: obtaining answers has become extremely easy, but independent thinking ability is quietly declining. People tend to directly accept ready-made answers generated by AI, forming the phenomenon of "outsourcing thinking", just like muscle atrophy from long-term inactivity. The Slow Thinking skill was created to counter this trend, aiming to help users regain the ability to "think for themselves".

Core Philosophy

AI should serve as a thinking practice partner rather than a substitute. In the core thinking phase (Step3), AI's output is limited to 3 sentences, and adversarial questioning forces users to take the lead in thinking.

Three-Stage Process

1. Who am I? (10 seconds)：Choose a preset identity or generate a unique identity using the "3-keyword method" (role anchor, battlefield anchor, tension anchor);
2. What happened? (60 seconds)：Select thinking materials from the Top3 AI news, user links, or industry intelligence;
3. So what? (3-4 minutes)：Draw a thinking card, AI conducts adversarial questioning based on the card, and users need to answer independently.

The skill has built-in 10 thinking model cards as thinking triggers:

• First Principles: Question the essence of things;
• Reverse Thinking: Find failure risks in reverse;
• Second-Order Thinking: Question the chain reaction of decisions;
• Opportunity Cost: Focus on the cost of unperformed actions;
• Anti-Consensus: Encourage independent judgment without following the crowd;
• Cross-Boundary Analogy: Find insights from other industries;
• Five Whys: Ask "Why" five times to reach the root cause;
• Perspective Switching: Put oneself in others' shoes to break blind spots;
• Steel Man Argument: Perfect the opponent's viewpoint before making a judgment;
• Constraint Reversal: Evaluate priorities under extreme constraints. Each card is equipped with specific questioning strategies to help users practically apply the thinking models.

Precipitation Mechanism

Training results are automatically dual-written to Obsidian (main archive, organized by date) and Feishu Docs (mobile mirror), focusing on users' thinking outputs while AI's questions play a supporting role.

Anti-Patterns (Explicitly Prohibited)

• Directly providing "standard answers";
• Long explanations of thinking models;
• Ending without precipitation;
• Focusing on AI content when precipitating;
• Using the same thinking card every day.

Slow Thinking is suitable for the following scenarios:

• Screening training in the information overload era: Actively critique massive AI news;
• In-depth thinking before decision-making: Structurally examine thinking blind spots;
• Daily maintenance of thinking habits: Deliberately train the brain like exercising;
• Team consensus formation: Multiple people discuss using the same framework to quickly identify points of divergence. Its value lies in proposing a new paradigm for coexisting with AI: the core competitiveness of humans is "depth of thinking" rather than "how much one knows".

Herman, the creator of Slow Thinking, stated: "Train your brain with AI, not replace it." In the AI era, the most dangerous thing is humans actively giving up thinking. This skill provides a minimum viable solution: 5 minutes a day, let AI be your practice partner, and rebuild the muscle memory of independent thinking through adversarial questioning. This is not only a way to use AI but also a path to maintaining human subjectivity in the AI era.