Taleb Skill

Why it's ranked here

The appeal is unusually concrete: the repository defines activation rules, a research workflow, failure checks, exit triggers, and worked conversations rather than offering a loose character prompt. Its source material also includes criticism and internal contradictions. The result looks useful for adversarial decision analysis, though the rigid persona limits broader advisory work.

What's good

The skill asks who bears a decision’s consequences, tests whether repeated exposure can cause ruin, and separates bounded experiments from destructive downside. For factual questions, it requires current research, opposing views, historical analogies, and concrete asymmetry. It also states where its own framework fails, including ordinary operational decisions and specialist domains.

Tradeoffs

Its prescribed voice is aggressive, highly certain, and explicitly hostile to balanced phrasing. That can sharpen analysis but distort uncertainty or damage sensitive conversations. The workflow depends on web-search capability for companies, markets, people, and current events. It admits that it is weak at concrete operating plans, gentle communication, and specialist judgments such as clinical medicine.

How to use it well

Use it as a red-team partner before investments, career bets, product commitments, or forecasts with irreversible downside. Give it the position size, worst case, incentives, and constraints. Pair its critique with domain experts and an implementation-oriented planner. Do not treat the performed persona as Taleb himself or as specialist financial or medical advice.

Observed

License

MIT License

Primary format

Markdown, with YAML frontmatter in the skill definition

Install surface

Install through the skills CLI, clone into a runtime skill directory, or paste the skill text into a conversation

Interface

Agent Skills-compatible prompt skill; no library or HTTP API is described

Platform support

Documented for skills-compatible runtimes including Claude Code, Codex, Cursor, OpenClaw, Hermes Agent, Gemini CLI, and OpenCode

Repository structure

Contains a skill definition, README, license, research references, and one demonstration conversation; no test directory appears in the supplied tree