tjboudreaux/cc-thinking-skills

What it solves

This project provides a structured way to improve how AI agents reason through complex problems. It addresses the need for clear, repeatable procedures when an agent is tasked with diagnosis, decision-making, strategy, or risk assessment, helping to avoid vague or unguided responses.

How it works

The catalog contains 28 "Agent Skills"—pre-defined reasoning frameworks like the Scientific Method, Five Whys, or Pre-Mortem. You can install these skills into compatible tools like Claude Code or GitHub Copilot. When you have a problem, you can either invoke a specific skill directly or use the included thinking-model-router to help the AI select the most appropriate framework for the task.

Who it’s for

It is designed for developers and users of AI agent tools who want to move beyond simple prompting and enforce rigorous, logical thinking patterns in their AI's output.

Highlights

• Includes 28 distinct reasoning frameworks covering diagnosis, strategy, and risk management.
• Compatible with Claude Code, GitHub Copilot, Codex, and Cursor.
• Features a "model router" to automatically suggest the best reasoning framework for a given problem.
• Modular design allows users to install specific skills or the entire library via CLI or manual file copying.
• Transparent documentation regarding performance, noting that these skills provide structured procedures rather than guaranteed accuracy improvements.