Skills that keep a person's own thinking in the loop when working with GenAI
概览
A collection of Agent/Claude Skills to help you think, instead of having your thinking taken from you by GenAI. These skills are designed to give you agency back. They ask questions, one at a time, and use your answers to decide what to ask next. In addition, these skills are intended to flag potential weaknesses that are inherent in all GenAI, such as sycophancy and verification errors. State the problem, or the plan - We solved this problem last quarter and it keeps coming back - Everyone sees a different cause for the same problem - We're considering bonuses for resolving client calls within five minutes - In three years we expect GenAI to do most of the work State the claim, or the decision - Adding another step in the review process will reduce the number of defects - I've decided to take a new job - My co-worker is getting under my skin and raising the problem will just make it worse - We selected a new supplier in ten minutes with no objections Share what the model...
README
以下为原文 README 节选，完整内容请查看原文。
Agent Thinking Skills
A collection of Agent/Claude Skills to help you think, instead of having your thinking taken from you by GenAI.
These skills are designed to give you agency back. They ask questions, one at a time, and use your answers to decide what to ask next.
In addition, these skills are intended to flag potential weaknesses that are inherent in all GenAI, such as sycophancy and verification errors.
The Skills
| Skill | What it’s for |
|---|---|
| Systems Thinking | Problems that keep coming back. Traces causes upstream, finds the feedback loops holding a situation in place, and draws Causal Loop Diagrams when the picture would help. |
| Critical Thinking | Claims, conclusions and decisions that need testing before you act. Paul and Elder’s eight Elements of Reasoning and nine Intellectual Standards, asked as questions - starting from wherever your thinking already is. |
| Critical Thinking for GenAI | Text a model produced, where asking the model to explain itself doesn’t work because it has no reasoning to retrace. Eleven probes that test the output itself. |
| Evaporating Cloud | Two reasonable choices that conflict. Draws Goldratt’s Evaporating Cloud to surface the assumptions behind each side. Probes for an action that meets both needs instead of splitting the difference. |
What To Ask
Systems Thinking - causes upstream, consequences downstream State the problem, or the plan
- We solved this problem last quarter and it keeps coming back
- Everyone sees a different cause for the same problem
- We’re considering bonuses for resolving client calls within five minutes
- In three years we expect GenAI to do most of the work
For more detail see: Systems Thinking
Critical Thinking - testing human reasoning and its implications State the claim, or the decision
- Adding another step in the review process will reduce the number of defects
- I’ve decided to take a new job
- My co-worker is getting under my skin and raising the problem will just make it worse
- We selected a new supplier in ten minutes with no objections
Critical Thinking for GenAI - testing output that has no human reasoning Share what the model generated
- It quoted a statistic without a source
- It summarised a sensitive meeting and agreed with claims people made in it
- I asked two different LLMs and they agreed
- It reviewed the contract and said “no flags”
Evaporating Cloud - conflicts that end in compromise State the two options, or the argument
- Promote the new CEO from within, or hire from outside
- Every planning session ends in the same fight over features versus technical debt
- Part of me wants to save for a trip, part of me wants to eat out more
- Two teams each need the same specialist full-time
Roadmap
推荐工具
换一个关键词，或者移除筛选条件。
安装
npx skillfish add mlevison/agent-thinking-skills