Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问许可证需确认审计通过

analogical-thinking类比思维

Agent Skill

analogical-thinking 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 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

523

周安装

22

GitHub Stars

6

下载量

183
CodexClaudeCursorGemini CLI

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

unknown

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:analogical-thinking(类比思维)
来源仓库:https://github.com/andurilcode/skills
仓库路径:skills/analogical-thinking
安装命令:
npx skills add https://github.com/andurilcode/skills --skill analogical-thinking
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 npx skills 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。

skills.shnpx skills
npx skills add https://github.com/andurilcode/skills --skill analogical-thinking

简介

类比思维技能聚焦识别问题深层结构与已有解决方案之间的映射关系,提升复杂问题解决效率。

  • 主张大多数难题都有历史先例,关键在于剥离表象差异,转移成熟方法论到新领域。
  • 借鉴 TCP 拥塞控制、进化算法等经典案例,强调结构相似性而非表面特征匹配。
  • 警惕虚假类比风险,需持续评估类比边界,避免因误判导致策略失效。
  • analogical-thinking 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Analogical Thinking

Core principle: Most genuinely hard problems have structural analogues elsewhere — often solved long ago under a different name. Recognize the *shape* of the problem beneath surface details, then transfer the solution structure.

The ctx harness as OS memory management. Blackboard pattern from speech recognition (1977) re-emerging in multi-agent AI. TCP congestion control inspiring rate-limiting. Evolution as search algorithm.

The risk: false analogies — surface similarity masking structural difference. This skill is as much about knowing when an analogy breaks as when it applies.


Core Process

Step 1: Abstract the Problem Structure

Strip domain vocabulary:

  • What needs to coordinate with what?
  • What needs to be stored, retrieved, prioritized, transformed, routed?
  • What's the flow? What are the constraints?
  • What failure modes are you preventing?

The more abstract, the wider the search space.

Example: *"How do agents share intermediate results without stepping on each other?"* → *"Multiple concurrent writers contribute partial results to a shared workspace, with coordination to prevent conflicts and allow selective reading."* Now sounds like distributed systems, DB concurrency, collaborative editing.

Step 2: Search for Structural Analogues

Look across:

  • Natural systems — evolution, immune response, neural networks, ant colonies, ecosystems, markets
  • Engineering — civil, mechanical, electrical, chemical (centuries of patterns)
  • CS classics — OS design, compiler theory, networking, DB internals, distributed systems
  • Organizational theory — military command, jazz improvisation, surgical teams, ATC
  • Biology — cell signaling, protein folding, predator-prey, homeostasis
  • Physics / Information theory — entropy, signal/noise, conservation laws, phase transitions

Step 3: Evaluate Strength

Useful when:

  • Relationships between components map cleanly (not just the components)
  • Constraints are similar in kind (even if not in degree)
  • Failure modes of the source are informative

Breaks when:

  • Key properties of the original don't hold here
  • Scale difference produces different emergent behavior
  • Analogy explains structure but not dynamics (or vice versa)

Always ask: *"Where does this analogy fail? What's different that matters?"*

Step 4: Transfer the Solution Pattern

  • What's the core mechanism (not implementation)?
  • What adaptations are needed?
  • What accumulated refinements can you inherit?

High-Value Source Domains

DomainSolvedUseful for
Operating SystemsResource allocation, memory management, scheduling, concurrency, isolation, caching, virtual addressingAgent orchestration, multi-tenancy, LLM context, pipelines
Distributed SystemsConsensus under failure, eventual consistency, partition tolerance, idempotency, log-structured storage, leader electionMulti-agent coordination, resilient pipelines, sync
Ecology / EvolutionAdaptation under selection, niche differentiation, resource competition, co-evolution, resilience via diversityAdversarial systems, red teams, org adaptation, market strategy
Control TheoryFeedback loops, stability, overshoot, damping, PID, observabilityMonitoring, auto-scaling, goal-seeking systems
Military / LogisticsCommand under uncertainty, supply chains, mission planning under partial info, combined armsIncident response, large-scale planning, agent coordination
Jazz / ImprovisationStructured improvisation, real-time coordination without central control, shared vocabulary enabling emergenceTeam autonomy with alignment, agents under ambiguity

Output Format

Abstracted Problem Structure

Re-stated in domain-neutral terms: core dynamic, key constraints, failure modes.

Structural Analogues Found

For each candidate:

  • Source domain: where the pattern comes from
  • Analogue structure: how the source solved it
  • Fit: Strong / Partial / Weak
  • Where it holds: specific correspondences
  • Where it breaks: key differences

Transferred Solution Pattern

  • Structure of the solution
  • Adaptations needed
  • Refinements worth inheriting
  • Failure modes to avoid (source domain learned the hard way)

False Analogy Risks

  • Surface similarity that suggests stronger fit than exists
  • Properties of the source that don't hold here

Classic Analogies in Software / AI

ProblemSourceWhat transferred
Agent context managementOS virtual memory + pagingActive context = RAM; long-term storage = disk; page faults = retrievals
Multi-agent coordinationBlackboard architecture (Hearsay-II, 1977)Shared workspace; specialists read/write; no direct communication
LLM token limitsCPU cache hierarchyWorking memory vs. storage; cache misses as retrievals
Agent pipelineScientific methodHypothesis → experiment → observation → update → repeat
Prompt compressionData compression / entropy codingLossless vs. lossy; semantic entropy as measure
AI red teamingSecurity pentestAdversarial mindset, kill chains, surface enumeration

Thinking Triggers

  • *"What's the shape of this problem, stripped of its domain vocabulary?"*
  • *"Who has solved a problem with this structure before, in any field?"*
  • *"Where does this analogy feel strong? Where strained?"*
  • *"What did the source domain learn the hard way that we can inherit for free?"*
  • *"If this were a networking problem / biology problem / physics problem, what would the answer look like?"*

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

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

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

Codex

38.53%
按下载量换算71

Claude

31.33%
按下载量换算57

Cursor

17.66%
按下载量换算32

Gemini CLI

8.93%
按下载量换算16

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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