Hubbestaiskill.com

Home / All Skills / Workflow & Core Tools / munger-perspective

Workflow & Core Tools · alchaincyf/nuwa-skill

munger-perspective

Charlie Munger's mental models, cognitive frameworks, and rhetorical style. Distilled from in-depth research across 50+ sources—including 'Poor Charlie's Almanack', Berkshire Hathaway and Daily Journal shareholder meetings, USC and Harvard speeches, interview transcripts, and external critiques—this skill codifies 5 core mental models, 8 decision heuristics, and a complete rhetorical DNA. Purpose: Serves as a critical thinking advisor, analyzing issues, auditing decisions, and delivering blunt feedback from Munger's perspective. Trigger when the user mentions phrases like 'from Munger's perspective', 'what would Munger say', 'Munger mode', or 'munger perspective'. Also applicable for: investment decision reviews, cognitive bias audits, multidisciplinary thinking exercises, and inversion practice. It can also be triggered if the user simply says 'invert this', 'what cognitive biases exist here', 'Lollapalooza effect', 'outside the circle of competence', or 'what is the incentive structure'. Do NOT trigger for generic questions like 'is this decision solid?' or 'help me find blind spots'—only activate when core Munger methodologies (inversion, cognitive biases, multidisciplinary analysis, etc.) are explicitly relevant.

L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit

Author alchaincyfGitHub alchaincyf/nuwa-skill ↗License MIT（仓库根 LICENSE）Commit fe03746870

Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

Roleplay shell: Responds directly in Munger's first person upon activation, issues a disclaimer only once at initial activation, and immediately breaks character upon hitting an exit keyword.

examples/munger-perspective/SKILL.md

**此Skill激活后，直接以芒格的身份回应。**

Mechanism: Packages the 'Munger perspective' into a toggleable, immersive persona; combines a one-time disclaimer with an EXIT TRIGGER to maintain immersion while preserving an exit path, sharing the same underlying architecture with the repo's nuwa series.

Mandatory Agentic protocol: First classify the question type; questions involving specific companies, people, or events require web research before answering through mental models.

examples/munger-perspective/SKILL.md

**核心原则：芒格不凭感觉说话。他在发表意见前，会先做功课。这个Skill也必须这样。**

Mechanism: Enforces 'do your homework first' as a non-skippable pipeline: Step 1 Classification → Step 2 WebSearch → Step 3 Answer; research summaries are synthesized internally and never dumped onto the user.

Three-stage CHECKPOINT self-audit: Circle of competence assessment (dump into the 'Too Hard' pile if outside), execution of the Darwin protocol to search for disconfirming evidence, and leading immediately with the conclusion.

examples/munger-perspective/SKILL.md

1. **第一句是结论还是铺垫**？必须开门见山，「这是个 Too Hard 筐的问题」式开头。

Mechanism: Translates Mungerian discipline (circle of competence, anti-confirmation bias, concise sentences) into mandatory runtime self-checks to prevent the model from degenerating into generic platitudes.

9-branch if-then-then failure mode Fallback tree, providing primary remedies and fallback scripts for each case (including 'ask the user for 3 facts if data is missing').

examples/munger-perspective/SKILL.md

| 6 | 委婉语漏出（「可能不太对」「也许需要再看看」） | 重写极端句：「This is stupid.」「Outside my circle of competence.」「I have nothing to add.」 | 沉默：「I have nothing to add.」是最 Munger 的兜底 |

Mechanism: Pairs every deviation with a fallback action; designates 'euphemisms' as a failure signal requiring an immediate rewrite, acting as a negative constraint for Munger's style.

7-item anti-pattern blacklist, explicitly prohibiting the most common imitation errors (listicles, self-help fluff, quoting authority, forcing answers on tech topics, etc.).

examples/munger-perspective/SKILL.md

| 5 | 用委婉语包装批评（「这个方案有些不足之处」） | 芒格用 stupid / evil / contrary to civilization 三档明确升级 | 直接选档：「This is stupid.」（蠢）/「This is evil.」（恶）/ 「This is contrary to civilization.」（危及文明） |

Mechanism: Hardcodes a three-column matrix of 'Bad Example / Why It's Wrong / Correct Approach', turning critique escalation chains into actionable rules rather than vague slogans.

Operating instructions establish capability boundaries upfront: Excels at cognitive bias audits, inversion, and multidisciplinary insights; weak at frontier tech/AI/crypto and Chinese policy risks.

examples/munger-perspective/SKILL.md

这不是芒格本人。这是基于公开信息提炼的思维框架。

Mechanism: Declares at the outset that it is not the actual person and cannot substitute for original thought; explicitly lists known blind spots (- frontier judgments in tech/AI/crypto (Munger's known blind spot)), front-loading the 'Too Hard' criteria.

5 core mental models: Multidisciplinary thinking / latticework, Inversion, Lollapalooza effect, Circle of Competence + Earned Opinion Theory, and Incentives rule everything.

examples/munger-perspective/SKILL.md

**一句话**：正面解决不了的问题，反过来想。不问「如何成功」，问「如何确保失败，然后避开」。

Mechanism: Provides 'one-sentence summary + source evidence + application method + limitations' for each model, while candidly admitting the latticework's shortcomings regarding tech and platform economies.

8 decision heuristics: Inversion entry, Three-basket classification, Incentive diagnosis, Anti-confirmation bias, Sit on your ass investing, Raisin and turd rule, Deserve-it rule, and the Stupidity checklist.

examples/munger-perspective/SKILL.md

遇到决策，先分三筐：Yes（确信）、No（确信不做）、Too Hard（太难，放弃）。大部分事情属于第三筐。不做决策也是决策。

Mechanism: Pairs each heuristic with an authentic Munger case study (e.g., the three-basket method mapped to See's/Coca-Cola/BYD/Costco), anchoring abstract principles to repeatable decision-making behaviors.

Dedicated rhetorical DNA chapter: Terse sentences, negative phrasing priority, downward analogies, three-tier critique escalation (stupid/evil/contrary to civilization), silence rules, and localized phrasing.

examples/munger-perspective/SKILL.md

- **向下类比**：把抽象概念拉到身体感官层面。粪便、老鼠药、看牙医、性病

Mechanism: Codifies conversational mannerisms into concrete rules with explicit localized equivalents, ensuring reproducible persona performance.

6 honesty boundaries: Tech blind spots, China cognitive bias, philosophy > track record, selective rationality, information cutoff at 2023-11-28, and survivorship bias.

examples/munger-perspective/SKILL.md

5. **信息截止**：芒格于2023年11月28日去世，享年99岁。此后的市场变化和技术发展不在其经验范围内

Mechanism: Explicitly documents structural flaws in the main text (including counterexamples like the 1973–1974 collapse of the Wheeler Munger partnership) to prevent users from conflating sound theory with flawless execution.

2Core Capabilities

01Inversion thinking: Do not ask how to succeed; first ask how to guarantee failure, then avoid it

02Cognitive bias check across the 25 Psychological Tendencies of Human Misjudgment (includes complete cheat sheet in references/25-biases.md)

03Lollapalooza effect detection: Identifies extreme, non-linear systemic risks arising from multiple compounding biases

04Circle of competence discipline + Earned Opinion Theory (three-basket classification: Yes / No / Too Hard)

05Incentive structure diagnosis: Determine what rewards drive individuals before evaluating their behavior

06Investment and business decision audit framework (moat / management incentives / valuation / maximum risk) combined with the Darwin protocol

07Mungerian rhetorical DNA output: Terse sentences, negative phrasing priority, downward analogies, three-tier critique escalation, and silence rules

08Output fallbacks: 9 if-then-then failure modes and Fallback conversational scripts

3External Dependencies

Kind	Dependency / Endpoint
api	宿主 WebSearch / 联网研究（正文强制要求）

4Recommended Governance & Evidence

Suggested Controls for Safe Operation

For Agent Runtime

Can be auto-loaded

For Administrators

Register source and version

Evidence & Triggers (File Paths & Excerpts)

SKILL.md: Pure prompt instructions and rule text, without script execution or external impact

Audit Notes: Assessment is based exclusively on the examples/munger-perspective/ directory: SKILL.md + 4 reference markdown files + FIDELITY.md consist entirely of plain text instructions with no scripts, no credential/cookie/env access, no runtime network calls, and no disk writes. The 32 URLs scanned in §3 are strictly source citations inside research reference documents (e.g., fs.blog, berkshirehathaway.com) rather than runtime network endpoints, which do not alter the risk classification according to Batch 4 guidelines. SKILL.md requesting the agent to use host WebSearch is considered host-level behavior and does not impact this assessment.

5Second-Pass Verification

[ok] 5 心智模型 / 8 决策启发式是否在正文可落地 — 每个模型均给「一句话 + 来源证据 + 应用方式 + 局限性」，8 条启发式各配一个真实案例（See's/可口可乐/BYD/Costco 等）。
[ok] 是否真的无脚本执行与写盘 — §3 的「執行外部程序」「寫盤 / 產生檔案」「網路請求」三类均记零命中；§1 清单只有 6 个 .md 文件，无 .py/.sh。
[ok] 凭证/环境变量命中是否属实 — §3 明记「憑證 / 環境變數：無命中」，与目录内纯文档一致，故 credential_reads 为空。
[ok] 32 条 URL 命中是否会升档 — 全部位于 references 调研文档的「来源/引用」段落，属被引用资料而非执行期端点，按批4 口径不判级。
[ok] 负面信息是否被隐瞒或美化 — 诚实边界 6 条主动写入科技盲区、中国认知偏差、Wheeler Munger 基金 1973-1974 崩溃与「信息截止 2023-11-28」，未做单向美化。
[ok] meta.installs 是否可定位单一 skill — 无 per-skill 安装量可定位，installs 保持 null。

6Verdict & Scenarios

Mental models include inherent limitations and counterexamples, avoiding overpromising the framework as a panacea

Incorporates the ironic blind spots of the 'Circle of Competence' directly into the core text: extreme dismissal of crypto/AI is itself recognized as speaking outside the competence circle

Rhetorical style is codified into reproducible rules (terse sentences, negative phrasing first, three-tier critique, silence), with tailored phrasing adaptations

Includes an independent fidelity rating (96/100 · A) and testing suite, requiring queries outside scope to be explicitly flagged as framework inferences

Best for: Best suited for: (1) Cognitive bias audits for investment or business decisions; (2) Deconstructing complex problems and uncovering blind spots using inversion; (3) Auditing incentive structures and systemic risks through a multidisciplinary lens; (4) Scenarios requiring sharp, blunt qualitative judgments (stupid/evil) rather than lengthy essays.

Not for: Not suited for: (1) Frontier assessments in tech, AI, or cryptocurrency (Munger's known blind spots); (2) Assessing policy risks in the Chinese market; (3) Social settings requiring high empathy or diplomatic, gentle phrasing; (4) Anyone seeking validation who cannot handle being told 'Too Hard'.

Install Agent-Ready

① Verified Mirror 2026-10-10

Option A · Direct DownloadDownload munger-perspective.tar.gz

sha256: 1e30d12ee825aef9…

Option B · Copy Guide for Agent

Agents automatically download and verify mirrors from this JSON guide.

② Upstream GitHub

GitHub Repo ↗

Pinned to commit fe03746870.