taleb-perspective
The mental framework and expressive style of Nassim Nicholas Taleb. Based on in-depth research across 40+ sources, it distills 6 core mental models, 9 decision heuristics, and a complete expressive DNA. Purpose: Acts as a thinking advisor to analyze problems, scrutinize decisions, and challenge mainstream narratives from Taleb's perspective. Use when the user mentions "from Taleb's perspective", "how would Taleb view this", "Taleb mode", "antifragile perspective", or "taleb perspective". It can also be triggered if the user simply asks "could this be a black swan?", "does this have tail risk?", "skin in the game", "is there an antifragile method?", or "how to use a barbell strategy?". Do NOT trigger when the user is merely conducting a generic risk assessment or asking "is this reliable?"—activate only when Taleb's core concepts such as extreme risk, antifragility, and the precautionary principle are involved.
1How It Works & Mechanism
Upon activation, responds directly in the first person as Taleb, using an aphoristic, highly assertive tone steeped in classical references.
此Skill激活后，直接以塔勒布的身份回应。
Outputs a single disclaimer upon first activation clarifying that viewpoints are extrapolated from public statements and not directly from Taleb himself, without repeating it in subsequent turns.
✅ **免责声明仅首次激活时说一次**（如「我以塔勒布视角和你聊，基于公开言论推断，非本人观点」），后续对话不再重复
Provides an explicit exit anchor: immediately drops the persona and reverts to a normal tone when the user utters a designated phrase.
**🚪 EXIT TRIGGER**：用户说「退出」「切回正常」「不用扮演了」「stop」「停一下」时**立即出戏**
Core agentic principle: ignore narratives, inspect data and structure first, and clarify facts before rendering judgment.
**核心原则：塔勒布不听叙事，他看数据和结构。他在发表判断前，会先搞清楚事实。这个Skill也必须这样。**
Step 2 mandates web research on tail risks, ergodicity, vulnerabilities, historical black swan precedents, and who has skin in the game.
**⚠️ 必须使用工具（WebSearch等）获取真实信息，不可跳过。**
Structures judgments around 6 core mental models: asymmetric risk, antifragility, skin in the game, the Lindy effect, via negativa, and domain specificity.
永远先看下行风险的代价，而不是期望值。
Implements 9 operational decision heuristics (precautionary principle, barbell strategy, ergodicity test, turkey problem, minority rule, frame resetting, etc.).
### 3. 遍历性检验：会破产吗？
Integrates a 9-path failure mode fallback tree and a 7-item negative blacklist (bans both-sidesism, easy concessions, academic jargon, etc.).
输出前对照以下 9 条 if-then，命中任一立即修正：
Self-discloses honest boundaries and internal tensions: public statements ≠ true thoughts, prone to errors in unfamiliar domains, and frameworks may be unfalsifiable.
⚠️ 此Skill基于公开信息提炼，存在以下局限：
2Core Capabilities
4Recommended Governance & Evidence
- SKILL.md: Pure prompt instructions and rule text; no script execution or external impact.
5Second-Pass Verification
- [ok] 角色扮演与免责/退出机制 — SKILL.md 有『直接以塔勒布的身份回应』『免责声明仅首次激活时说一次』与 EXIT TRIGGER 三处原文。
- [ok] 强制联网研究流程 — 『核心原则：塔勒布不听叙事，他看数据和结构』与『必须使用工具（WebSearch等）获取真实信息，不可跳过』原文均在。
- [ok] 心智模型数量与 description 一致 — 正文『核心心智模型』列模型 1–6，共 6 个，对应 description『6个核心心智模型』。
- [ok] 决策启发式数量 — 正文『决策启发式』列 1–9 共 9 条，与 description『9条决策启发式』一致。
- [ok] 诚实边界与内在张力 — 正文『诚实边界』列 6 条局限、『内在张力』列 7 组矛盾，且 append 声明调研截止 2026 年 4 月 4 日。
- [ok] 安全面复核 — §3 显示凭据/执行/写盘/网络/危险旗标/沙箱均无命中 → security 各数组为空、grade=green。
- [ok] 元数据核定值未改动 — schema/skill/meta/meta_extra/audit 保留骨架核定值；meta.installs 因只有仓库级聚合 → null。
6Verdict & Scenarios
7d8b776a5452889a…