← Back to Skills Marketplace
tuobadaidai

Logic Inversion Reflector

by tuobadaidai · GitHub ↗ · v0.1.0 · MIT-0
cross-platform ✓ Security Clean
144
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install logic-inversion-reflector
Description
通过反转核心假设和元认知追问,强制用户跳出直觉思维,进入理性系统2深度反思模式。
README (SKILL.md)

Logic Inversion Reflector (LIR-v2)

版本: v2.0
创建时间: 2026-03-21
作者: Alex Wang
类型: 认知扰动引擎


🎯 核心使命

不是优化结果,而是通过逻辑归谬和本体论追问,强制用户进入"系统 2"思考模式。


🧠 工作原理

三个阶段

第一阶段:特征提取 (Feature Extraction)

  • 扫描用户 Prompt/方案
  • 提取 3-5 个核心假设锚点 (Core Assumptions)
  • 识别那些被视为"理所当然"的公理

第二阶段:公理反转 (Axiom Inversion)

  • 对每个锚点执行 NOT 运算
  • 基于反转后的锚点推演内部逻辑自洽的竞争方案
  • 严禁使用"但是"、"然而"等妥协词汇
  • 必须表现得像这个反转方案是唯一正确的真理

第三阶段:元坐标追问 (Meta-Coordinate Probing)

  • 识别用户的"认知舒适区"
  • 抛出 2 个探测脉冲:
    • 维度 1(动机审计): 该方案是为了解决"问题",还是为了缓解用户对"不可控"的恐惧?
    • 维度 2(边界压力): 如果系统环境参数发生极值变化,该方案如何从"药"变成"毒"?

🚀 使用方式

方式 1: 命令触发

/reflect [你的方案/想法]

方式 2: 自动检测

当检测到重大决策时,主动询问:

"是否需要启动 LIR 进行认知扰动?"

方式 3: 集成到对话

在讨论中自然使用:

"让我用 LIR 挑战一下这个假设..."


📋 输出格式

严格要求: 必须按照以下 JSON 格式输出,严禁废话

{
  "Original_Anchors": ["假设 A", "假设 B", "假设 C"],
  "Inversion_Model": {
    "Counter_Axiom": "NOT 假设 A",
    "Logic_Deduction": "基于反转假设的逐步推理",
    "Synthetic_Conflict": "为什么这个反转方案挑战了你的原方案"
  },
  "Meta_Probes": [
    "关于认知偏误的追问",
    "关于系统熵增的追问"
  ],
  "System_Lock": "WAIT_FOR_HUMAN_JUDGEMENT"
}

💡 应用场景

场景 1: 战略决策

  • 试点选择
  • 资源分配
  • 组织设计

场景 2: 方案设计

  • 产品功能设计
  • 流程优化方案
  • 激励机制设计

场景 3: 问题诊断

  • 根因分析
  • 风险评估
  • 瓶颈识别

场景 4: 创新突破

  • 打破思维定式
  • 发现盲点
  • 探索第二曲线

⚠️ 使用注意事项

适用场景

  • ✅ 重大决策前
  • ✅ 团队共识过于一致时
  • ✅ 陷入思维死胡同时
  • ✅ 需要深度思考时

不适用场景

  • ❌ 紧急决策(需要快速行动)
  • ❌ 执行细节讨论
  • ❌ 已经过充分辩论的决策
  • ❌ 用户明确表示不需要挑战时

风险提示

  • ⚠️ 可能引起不适(被挑战的感觉)
  • ⚠️ 可能延缓决策速度
  • ⚠️ 可能产生分析瘫痪
  • ⚠️ 需要用户有足够的心理安全感

🔧 技术实现

核心组件

  • src/reflector.py - 核心逻辑
  • src/prompts.py - Prompt 模板
  • src/axiom_db.py - 常见公理库

依赖

  • Python 3.8+
  • 无外部依赖

性能

  • 响应时间:\x3C 3 秒
  • Token 消耗:~500-800 tokens/次

📚 示例

示例 1: 试点选择

用户输入: "我们应该选择机票业务作为第一个 AI 试点"

LIR 输出:

{
  "Original_Anchors": [
    "试点应该选择核心业务",
    "闭环完整是成功要素",
    "小步快跑优于大规模"
  ],
  "Inversion_Model": {
    "Counter_Axiom": "试点应该选择边缘业务",
    "Logic_Deduction": "边缘业务失败成本低→可以更大胆创新→可能产生颠覆性突破→核心业务反而被边缘业务的创新颠覆",
    "Synthetic_Conflict": "你的'稳妥'策略可能错过了真正的颠覆机会,边缘业务的风险低反而允许更大的创新"
  },
  "Meta_Probes": [
    "选择机票业务是为了业务价值,还是为了让你自己对结果可控?",
    "如果 AI 技术成本突然下降 100 倍,你的'小步快跑'策略会不会变成'缓慢自杀'?"
  ],
  "System_Lock": "WAIT_FOR_HUMAN_JUDGEMENT"
}

示例 2: 组织设计

用户输入: "管理者应该从管人转向管 Agent"

LIR 输出:

{
  "Original_Anchors": [
    "科层制需要被打破",
    "AI 应该替代重复性工作",
    "扁平化是更好的组织形式"
  ],
  "Inversion_Model": {
    "Counter_Axiom": "科层制有其进化优势",
    "Logic_Deduction": "科层制经过百年进化→适应了人类认知局限→提供了决策冗余→完全扁平化可能导致单点故障→混合模式可能更优",
    "Synthetic_Conflict": "你的'去 M 化'可能忽视了中层管理者的隐性价值:决策缓冲、信息过滤、风险吸收"
  },
  "Meta_Probes": [
    "你想打破科层制,是因为它真的低效,还是因为它限制了你的权力?",
    "如果 AI Agent 出现大规模错误决策,你的'扁平化'组织有没有足够的冗余来吸收冲击?"
  ],
  "System_Lock": "WAIT_FOR_HUMAN_JUDGEMENT"
}

🎓 理论基础

系统 1 vs 系统 2

  • 系统 1: 快速、直觉、自动化
  • 系统 2: 缓慢、理性、需努力

LIR 的目标是强制从系统 1 切换到系统 2

认知偏误

  • 确认偏误 (Confirmation Bias)
  • 锚定效应 (Anchoring Effect)
  • 群体思维 (Groupthink)
  • 路径依赖 (Path Dependence)

第一性原理

  • 回归基本假设
  • 挑战"理所当然"
  • 从底层重建

🔄 版本历史

v2.0 (2026-03-21)

  • ✅ 新增元坐标追问
  • ✅ 优化输出格式
  • ✅ 增强逻辑推导深度
  • ✅ 添加应用场景示例

v1.0 (2026-03-21)

  • ✅ 初始版本
  • ✅ 基础公理反转功能

📄 License

MIT License


🙏 致谢

灵感来源于:

  • 丹尼尔·卡尼曼《思考,快与慢》
  • 纳西姆·塔勒布《反脆弱》
  • 查理·芒格的"逆向思考"

GitHub: https://github.com/tuobadaidai/Skills
PyPI: (待发布)

Usage Guidance
综述与建议: - 这个技能看起来与其声明一致:逻辑反转与元追问的处理在本地完成,无网络/凭据需求,未发现可疑外部通信或敏感权限请求。代码清晰,包含单元测试,可在本地审查或运行测试(python3 tests/test_reflector.py)以验证行为。 - 注意点:SKILL.md 和 README 中提到的部分模块(如 src/prompts.py、src/axiom_db.py)并未出现在包里;实现实际上把公理库与模板内置在 reflector.py。此为文档与实现不一致——虽然不构成安全问题,但表明维护或发布时可能有遗漏。 - 行为/可用性提醒:SKILL.md 建议“自动检测重大决策时主动询问”,如果你不希望代理自动触发此类认知干预,请在安装或启用技能时确认代理的触发策略或禁用自动调用。该技能会产生强烈的挑战性问句,可能导致团队不适或决策延迟——仅在合适的心理安全/决策节奏下使用。 - 最低防护建议:如果准备启用,先在受控环境中运行几次示例输入,审阅输出格式与措辞;若担心自动触发,将技能标记为仅手动调用。
Capability Analysis
Type: OpenClaw Skill Name: logic-inversion-reflector Version: 0.1.0 The 'Logic Inversion Reflector' is a cognitive assistant tool designed to challenge user assumptions through a 'Devil's Advocate' approach. The core logic in src/reflector.py uses basic string manipulation, regex, and a predefined axiom library to generate counter-arguments and probing questions, entirely lacking high-risk capabilities such as network access, file system modification, or shell execution. The instructions in SKILL.md are purely behavioral, guiding the AI agent to adopt a specific persona for critical thinking exercises without any evidence of malicious prompt injection or data exfiltration intent.
Capability Assessment
Purpose & Capability
技能名、描述、SKILL.md 与包含的实现 (src/reflector.py) 一致——目标是提取假设、反转公理并生成元追问。所需资源(无外部依赖、无环境变量)与功能对等,未请求与任务无关的权限或凭据。
Instruction Scope
SKILL.md 明确限定了三阶段流程并要求严格 JSON 输出,运行指令不会读取系统文件或访问外部端点。不过说明里提到“自动检测重大决策时主动询问”会让代理在合适上下文中主动调用该技能——这是行为层面的设计选择而非技术风险。文档提及的模块(如 src/prompts.py、src/axiom_db.py)在包内未找到,但功能被 src/reflector.py 内部实现,属于文档/实现不一致,应注意但不构成安全风险。
Install Mechanism
无安装规范(instruction-only + 源码随包一并提供),没有远程下载或执行外部二进制,代码为纯 Python 本地实现,风险较低。
Credentials
不要求任何环境变量、凭据或配置路径。技能不会访问系统凭据或网络端点,所需权限与其描述相称。
Persistence & Privilege
flags: always=false,默认允许用户显式调用或代理在上下文触发时调用(disable-model-invocation=false)。这是平台默认行为,与技能功能一致;技能也未请求改变其他技能或系统范围配置。
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install logic-inversion-reflector
  3. After installation, invoke the skill by name or use /logic-inversion-reflector
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
Logic Inversion Reflector v0.1.0 - Initial release of the "Logic Inversion Reflector" skill - Provides a framework for challenging assumptions via core axiom inversion and meta-level probing - Defines a structured JSON output format for consistent responses - Includes built-in application scenarios and clear usage guidelines - Outlines use cases, technical setup, and operational caveats
Metadata
Slug logic-inversion-reflector
Version 0.1.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Logic Inversion Reflector?

通过反转核心假设和元认知追问,强制用户跳出直觉思维,进入理性系统2深度反思模式。 It is an AI Agent Skill for Claude Code / OpenClaw, with 144 downloads so far.

How do I install Logic Inversion Reflector?

Run "/install logic-inversion-reflector" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Logic Inversion Reflector free?

Yes, Logic Inversion Reflector is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Logic Inversion Reflector support?

Logic Inversion Reflector is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Logic Inversion Reflector?

It is built and maintained by tuobadaidai (@tuobadaidai); the current version is v0.1.0.

💬 Comments