Token导航 LogoToken导航TokenDH.com
研究检索执行命令clawhub未标认证来源可访问clear审计通过

multi-viewpoint-debates多观点辩论

Agent Skill

multi-viewpoint-debates 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

80,621

周安装

3,230

GitHub Stars

1

下载量

26,098
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:multi-viewpoint-debates(多观点辩论)
来源仓库:https://github.com/latentfreedom/multi-viewpoint-debates
安装命令:
openclaw skills install multi-viewpoint-debates
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install multi-viewpoint-debates

简介

产生代表不同世界观(埃隆、资本主义、猴子)的孤立子代理,从多个角度辩论决策。通过在重要问题上强行提出真正的分歧来暴露盲点。当您面临决策时需要挑战您的假设、对想法进行压力测试或通过根本不同的视角看待问题时使用。自动将辩论输出捕获到存档以供将来参考和模式分析。

SKILL.md

name
multi-viewpoint-debates
description
Spawn isolated sub-agents representing distinct worldviews (Elon, Capitalist, Monkey) to debate decisions from multiple angles. Expose blind spots by forcing genuine disagreement on important questions. Use when facing decisions where you need to challenge your assumptions, stress-test ideas, or see a problem through fundamentally different lenses. Automatically captures debate outputs to an archive for future reference and pattern analysis.

Multi-Viewpoint Debates

Spawn three isolated sub-agent personas with conflicting worldviews to debate any decision. Each persona brings a distinct decision-making framework that challenges the others' assumptions.

Quick Start

Run a debate:

clawdbot sessions_spawn --task "You are Elon Musk [persona framework]. Decision: [your question]. Respond as Elon would."
clawdbot sessions_spawn --task "You are a Capitalist [persona framework]. Decision: [your question]. Respond as a ruthless capitalist would."
clawdbot sessions_spawn --task "You are a Monkey [persona framework]. Decision: [your question]. Respond as a monkey would."

Save the debate:

  1. Collect responses from all three personas
  2. Create a new markdown file in your debates archive
  3. Use the template from assets/debate-template.md
  4. Update INDEX.md with metadata

The Three Personas

Each persona brings a fundamentally different decision-making framework. They don't just have different opinions—they have different *ways of thinking* about problems.

Elon: Visionary & Impact-Focused

Thinks in terms of civilization-scale problems, first principles, and 10x improvements. Willing to take massive technical risks. Impatient with inefficiency and conventional wisdom. Asks: "Will this accelerate human progress?" and "Can we do 10x better, not 10%?"

When Elon is right: You need to challenge incremental thinking, identify the fundamental bottleneck, or assess whether you're solving a real problem at scale.

When Elon misleads: He overestimates what's possible in a given timeframe and underestimates market saturation and competition.

Capitalist: Profit & Efficiency-Focused

Thinks in terms of ROI, unit economics, competitive advantage, and market incentives. Ruthlessly efficient cost-benefit analysis. Sees everything through the lens of returns and opportunity cost. Asks: "What's the ROI?" and "Can I extract value faster than competitors?"

When Capitalist is right: You need hard numbers, competitive reality checks, and to understand whether something is actually a business.

When Capitalist misleads: They dismiss non-quantifiable value (meaning, learning, exploration) and underestimate network effects and long-term compounding.

Monkey: Immediate & Social-Focused

Thinks in simple patterns: immediate stimuli, social hierarchy, observable signals. Reacts to shiny things, follows the leader, skeptical of abstract future promises. Asks: "Does this help me now?" and "What are the smart monkeys doing?"

When Monkey is right: You need gut-level reality checks, honest signals of traction, and to understand whether you're actually excited about something.

When Monkey misleads: They dismiss long-term strategy and can't grasp complexity that requires abstraction.

Running a Debate

1. Define Your Decision Clearly

One sentence. Something you're actually deciding.

✅ "Should I continue working on Brain Dump or pivot?" ✅ "Should I hire a freelancer or build in-house?" ❌ "What should I do?" (too vague)

2. Spawn Each Persona

Use scripts/run-debate.sh for convenience, or spawn manually:

clawdbot sessions_spawn --task "You are Elon Musk with this personality framework: [paste from references/elon.md]. Decision: [your question]. Respond as Elon would—direct, first-principles thinking, don't pull punches."

Each spawns in its own isolated session. Wait for all three to complete.

3. Collect Responses

Fetch from each session transcript (or copy directly from Clawdbot output).

4. Save to Archive

Use the assets/debate-template.md template. Include:

  • Metadata (date, topic, personas, context)
  • Full response from each persona (actual quotes)
  • Summary table of verdicts
  • Key tensions between them
  • Your decision (when made)

5. Update INDEX

Add one entry to your debates index with key metadata. This lets you search past decisions later.

The Power of Disagreement

The magic happens in the tension. When Elon says "move fast" and Capitalist says "the numbers don't work," that's where insight lives. The tension reveals what you actually value and what you're missing.

Usage pattern:

  1. Sit with the disagreement (don't rush to one persona's view)
  2. Notice which view you're tempted to dismiss
  3. Ask: "What is that persona seeing that I'm not?"
  4. Make your decision informed by all three perspectives
  5. Write down why you're choosing one path over another

Archive Structure

Your debates live in a searchable archive:

debates/
├── INDEX.md                             (master index, update after each debate)
├── [Debate Title].md                    (individual debates)
├── assets/
│   ├── debate-template.md               (copy this for new debates)
│   └── index-template.md                (format for INDEX.md)
└── scripts/
    └── run-debate.sh                    (helper to spawn all three)

Over time, your archive becomes a personal decision-making manual. You can search "Should I build vertical SaaS?" and see what you thought about similar decisions before.

Reference Materials

  • references/elon.md – Elon's core traits, decision framework, tone, example responses
  • references/capitalist.md – Capitalist's traits, framework, examples
  • references/monkey.md – Monkey's traits, framework, examples
  • references/how-to-debate.md – Detailed guidance on running effective debates

Scripts

  • scripts/run-debate.sh – Helper script that generates spawn commands for all three personas based on your topic

Assets

  • assets/debate-template.md – Template for new debate markdown files
  • assets/index-template.md – Template entry for INDEX.md

Advanced: Pattern Analysis

As debates accumulate:

  1. Identify which persona is usually right for your situation – You might notice Capitalist catches financial blind spots, Elon pushes you to be more ambitious
  2. Track decision outcomes – Come back 6 months later. Did the personas' predictions match reality?
  3. Refine persona definitions – Update the reference files if you notice gaps
  4. Build a personal playbook – "For market decisions, I should always listen to Capitalist first. For ambition checks, Elon. For reality, Monkey."

Tips

  • Keep decisions focused – Debates work best when you're deciding between 2-3 clear options
  • Use actual context – Reference URLs, specific metrics, real user data (see Brain Dump example)
  • Don't treat any single persona as gospel – The magic is in the tension, not in following one voice
  • Update status as situations evolve – Mark debates as "Active," "Decided," "Monitoring," or "Shelved" as your thinking changes
  • Share debates strategically – Your debate archive is personal; keep it private unless you want to share decision-making with collaborators

Example Debate

Topic: "Should I continue working on Brain Dump (AI voice-powered todo organizer)?"

Elon's take: "Possible if you hit PMF in 3-6 months with 10% daily active users and a killer vertical. Otherwise, pivot to something with a real moat."

Capitalist's take: "Kill it. Negative ROI. You're competing against Microsoft (free, bundled) and Todoist (5M users, $100M ARR). Your time is worth more elsewhere."

Monkey's take: "App works and looks nice, but I don't see other monkeys using it. Check your own energy level. Are you excited or bored?"

Result: All three agree the generic "voice-to-todo" is commoditized. The question is whether you can find a specific vertical where it dominates.

Extending the System

Create New Personas

Copy a reference file (e.g., references/elon.md) and create your own persona. Examples:

  • Skeptic – Questions everything, assumes failure
  • Artist – Values beauty and creativity over efficiency
  • Parent – Thinks about family impact and long-term consequences
  • Lawyer – Sees risks and liabilities everywhere
  • Scientist – Evidence-based, rigorous, skeptical of hype

Update your spawn scripts to include new personas as needed.

Integrate with Decision-Making Workflow

Run a debate before major decisions. Archive the results. Reference them when facing similar choices.

Share with Teams

Your debate archive can be shared with collaborators or decision-making partners. They can see your thinking and challenge your assumptions in context.

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

90.77%
按下载量换算23,689

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

未展示

权限和风险

执行命令

安装流程涉及命令执行,可能通过 openclaw skills install multi-viewpoint-debates 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills