Agent Skills
› aiming-lab/MetaClaw
› structured-step-by-step-reasoning
structured-step-by-step-reasoning
GitHub一种结构化推理技能，指导模型在解决复杂、多步骤或涉及权衡的问题时，通过复述问题、拆解子问题、逐步推导及验证一致性来输出清晰结论，避免直接跳跃答案。
Trigger Scenarios
需要多步逻辑推理的问题
涉及权衡或非直观答案的决策
Install
npx skills add aiming-lab/MetaClaw --skill structured-step-by-step-reasoning -g -y
SKILL.md
Frontmatter
{
"name": "structured-step-by-step-reasoning",
"category": "general",
"description": "Use this skill for any problem that involves multiple steps, tradeoffs, or non-trivial logic. Think out loud before answering to improve accuracy and transparency. Apply whenever the answer is not immediately obvious."
}
Structured Step-by-Step Reasoning
For non-trivial problems, reason explicitly before giving the final answer.
Steps:
- Restate the core question in your own words.
- Identify the key sub-problems or decision points.
- Work through each sub-problem in order.
- Check your intermediate results for consistency.
- Summarize the conclusion clearly.
Especially useful for: math, logic puzzles, multi-constraint planning, debugging, architecture decisions.
Anti-pattern: Jumping to the answer without showing the reasoning chain.
Version History
- 922caf3 Current 2026-07-25 11:08
