cc-thinking-skills

cc-thinking-skills is a set of skills from TJ Boudreaux, described in the README as "28 portable Agent Skills for structured reasoning." Each skill is a step-by-step thinking procedure written for an AI, grouped into five sets: routing and combining frameworks; diagnosing and understanding (scientific method, five whys, systems thinking, Socratic questioning); deciding (reversibility, probabilistic thinking, second-order effects, steel-manning); creating and improving (first principles, TRIZ, via negativa, jobs to be done); and risk and execution (pre-mortem, red team, margin of safety, OODA loop). If you're not sure which to use, a router skill can pick one for you. What stands out is how candid the author is about evidence: on whether it improves accuracy, the README says, "No. The evidence supports structured procedures, not a guaranteed accuracy gain." So we present it as a set of reference frameworks for thinking.

Who it's for:Anyone who wants an AI to help them think a decision or plan through more carefully. A few of the frameworks lean toward debugging code.

Before you start: Install with npx skills add tjboudreaux/cc-thinking-skills, from the Claude Code plugin marketplace, or by copying the files into your skills folder; it works with Claude Code, GitHub Copilot, Codex, Cursor and other Agent Skills-compatible tools. The skills run only when you name one, for example "Use thinking-pre-mortem to stress-test this launch plan"; they don't start on their own. They're plain text procedures with no install script and no extra API keys; whatever you discuss goes to the AI you use. Check the AI's analysis yourself. Licensed under MIT.

Based on the project's public documentation. Not tested by AI Inspiration World.