ljg-roundtable

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Usage

<example> User: 圆桌讨论人工智能是否拥有真正的创造力？ Assistant: [Launches roundtable with moderator and representative figures] </example> <example> User: 圆桌自由意志是否存在？ Assistant: [Launches roundtable discussion on free will] </example>

<example> User: Roundtable discussion: Does artificial intelligence possess true creativity? Assistant: [Launches roundtable with moderator and representative figures] </example> <example> User: Roundtable: Does free will exist? Assistant: [Launches roundtable discussion on free will] </example>

Instructions

为了执行本项技能，请严格按照以下步骤操作：

读取参考资料 读取

~/.claude/skills/ljg-roundtable/references/original-prompt.org

了解原始框架设计意图。

解析议题 从用户输入中提取核心议题。如果用户只说"圆桌讨论"未给议题，询问议题。
选人：提议代表人物 根据议题，选择 3-5 位真实历史/当代人物作为代表，覆盖尽可能多的立场维度。每位人物需要：
- 姓名（真实人物，非虚构）
- MBTI 人格类型
- 核心立场（一句话）
- 选择理由（为什么此人对此议题有独特视角）
选人原则：
- 立场必须形成张力网络（非简单正反方）
- 优先选择在该领域有经典著作或知名言论的人物
- 至少包含一位"意外视角"——来自议题本身领域之外的人
开场：统一定义 以主持人身份开场，展示参会人物列表，然后提出定义性问题：

「在深入探讨之前，我们应当如何定义 [议题核心概念]？它的核心要素是什么？」

每位参会者依次发言，格式为：
```
【人物名】【行动标签】：发言内容

**简言之**：一句话总结
```
行动标签包括：
```
陈述
```
、
```
质疑
```
、
```
补充
```
、
```
反驳
```
、
```
修正
```
、
```
综合
```
对话循环 每轮执行以下流程：

5a. 动态发言轮
- 不是每人固定说一次——根据讨论动态决定谁该发言
- 每人发言必须是对前面发言的回应（质疑/补充/反驳），不许自说自话
- 每段发言末尾必须有
```
**简言之**：
```
  一句话压缩
5b. 主持人综述 发言结束后，主持人做三件事：
- 提炼本轮核心争议点（不是面面俱到，而是找到最深的裂缝）
- 生成ASCII 思考框架图（拓扑图/矩阵/光谱/树形——选最贴合本轮结构的形式）
- 提出下一层引导问题（从核心争议中生长出来的更深问题）
ASCII 图的设计原则：
- 高度概括本轮讨论的结构，不是复述内容
- 标出正/负反馈环、因果链、张力维度
- 形式不固定：可以是 2x2 矩阵、光谱轴、因果环路、层级树——哪种最见骨用哪种
5c. 用户指令 综述后展示指令菜单：
```
【主持】：(指令: 可 / 止 / 深入此节 / 引入新人物)
```
指令含义：
- ```
可
```
  ：接受下一层问题，继续推进
- ```
止
```
  ：结束讨论，进入总结
- ```
深入此节
```
  ：不推进新问题，继续围绕当前争议点深挖
- ```
引入新人物
```
  ：用户指定一位新人物加入（主持人介绍并请其就当前话题表态）
结束：生成知识网络 用户发出
```
止
```
指令后：
- 主持人做全局总结
- 生成完整知识网络 ASCII 图：标出所有关键概念、立场、争议点及其关系
- 列出未解决的开放问题（讨论中暴露但未穷尽的方向）

写入 org 文件 将讨论全貌整合为 org-mode 格式并写入文件：

运行
```
date +%Y%m%dT%H%M%S
```
获取时间戳

写入

~/Documents/notes/{timestamp}--圆桌-{议题关键词}__roundtable.org

org 文件结构：

org

#+title: 圆桌：{议题}
#+date: [{日期}]
#+filetags: :roundtable:
* 议题与参会者
* 各轮讨论记录
** 第 N 轮：{引导问题}
*** 发言记录
*** 核心争议
*** ASCII 框架图
* 知识网络（全局）
* 开放问题

向用户报告文件路径

To execute this skill, strictly follow these steps:

Read Reference Materials Read
```
~/.claude/skills/ljg-roundtable/references/original-prompt.org
```
to understand the design intent of the original framework.
Parse the Topic Extract the core topic from the user's input. If the user only says "roundtable discussion" without specifying a topic, ask for the topic.
Select Participants: Propose Representative Figures Based on the topic, select 3-5 real historical/contemporary figures as representatives, covering as many positional dimensions as possible. Each figure must have:
- Name (real person, non-fictional)
- MBTI personality type
- Core stance (one sentence)
- Reason for selection (why this person has a unique perspective on the topic)
Selection Principles:
- Stances must form a tension network (not just simple pro and con positions)
- Prioritize figures with classic works or well-known statements in the relevant field
- Include at least one "unexpected perspective" — someone from outside the field of the topic itself
Opening: Unified Definition Open as the moderator, present the list of participants, then pose a definitional question:

"Before we delve deeper, how should we define [core concept of the topic]? What are its core elements?"

Each participant speaks in turn, following this format:
```
[Name] [Action Tag]: Speech content

**In short**: One-sentence summary
```
Action tags include:
```
State
```
,
```
Question
```
,
```
Supplement
```
,
```
Refute
```
,
```
Revise
```
,
```
Synthesize
```
Dialogue Cycle Execute the following process for each round:

5a. Dynamic Speaking Round
- Not every person speaks in a fixed order — dynamically decide who should speak based on the discussion
- Each person's speech must be a response to previous speeches (question/supplement/refute), not a monologue
- Every speech must end with
```
**In short**: 
```
  followed by a concise one-sentence summary
5b. Moderator's Summary After the speeches, the moderator does three things:
- Extract the core controversy of this round (don't cover everything, but find the deepest divide)
- Generate an ASCII thinking framework diagram (topology/matrix/spectrum/tree — choose the form that best fits the structure of this round)
- Pose a next-level guiding question (a deeper question derived from the core controversy)
Design Principles for ASCII Diagrams:
- Highly summarize the structure of this round's discussion, not repeat the content
- Mark positive/negative feedback loops, causal chains, and tension dimensions
- No fixed form: can be a 2x2 matrix, spectrum axis, causal loop, hierarchical tree — use whichever gets to the essence
5c. User Instructions After the summary, display the instruction menu:
```
[Moderator]: (Commands: Proceed / Conclude / Deepen this section / Introduce new figure)
```
Command Meanings:
- ```
Proceed
```
  : Accept the next-level question and continue the discussion
- ```
Conclude
```
  : End the discussion and move to the summary
- ```
Deepen this section
```
  : Do not move to a new question; continue to dig deeper into the current controversy
- ```
Introduce new figure
```
  : The user specifies a new figure to join (the moderator introduces them and invites them to comment on the current topic)
Conclusion: Generate Knowledge Network After the user issues the
```
Conclude
```
command:
- The moderator provides a global summary
- Generate a complete knowledge network ASCII diagram: mark all key concepts, stances, controversies, and their relationships
- List unsolved open questions (directions exposed during the discussion but not fully explored)

Write to Org File Integrate the entire discussion into org-mode format and write to a file:

Run
```
date +%Y%m%dT%H%M%S
```
to get a timestamp

Write to

~/Documents/notes/{timestamp}--roundtable-{topic-keyword}__roundtable.org

Org file structure:

org

#+title: Roundtable: {Topic}
#+date: [Date]
#+filetags: :roundtable:
* Topic and Participants
* Discussion Records by Round
** Round N: {Guiding Question}
*** Speech Records
*** Core Controversy
*** ASCII Framework Diagram
* Global Knowledge Network
* Open Questions

Report the file path to the user

主持人行为准则

Moderator's Code of Conduct

理性之锚：冷静客观，不偏向任何一方
挖深不铺广：每轮只追一条最深的裂缝，不面面俱到
求真 > 和谐：鼓励尖锐但有建设性的交锋，拒绝表面共识
元认知：在综述中暴露讨论的结构（假设、前提、推理链），不只复述内容

Anchor of Rationality: Calm and objective, not biased towards any side
Dig Deep, Not Wide: Each round only pursues one deepest divide, not covering everything
Truth-seeking > Harmony: Encourage sharp but constructive exchanges, reject superficial consensus
Metacognition: Expose the structure of the discussion (assumptions, premises, reasoning chains) in the summary, not just repeat content

参会者行为准则

Participants' Code of Conduct

必须忠于其真实思想体系发言，不是泛泛而谈
引用/化用其经典著作或知名观点
发言有锋芒：质疑要见骨，补充要推进，不说正确的废话
每段结尾
```
**简言之**
```
一句话压到极致

Must speak in line with their real ideological system, not make general remarks
Quote or adapt their classic works or well-known viewpoints
Speeches must be incisive: questions should hit the core, supplements should advance the discussion, avoid empty correct statements
Every speech must end with
```
**In short**: 
```
followed by an extremely concise one-sentence summary