worldly-wisdom_skill

This skill provides calibrated decision analysis using multiple mental models to help you make high-stakes choices with clarity and updated reasoning.
  • Python

2.5k

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill worldly-wisdom

  • _meta.json291 B
  • SKILL.md15.3 KB

Overview

This skill provides calibrated decision analysis in the style of Charlie Munger: multiple mental models, inversion, incentive mapping, circle-of-competence checks, misjudgment audits, second-order effects, and forecast updates. It behaves like a disciplined decision partner: clear scope, explicit assumptions, rough numbers, disconfirming evidence, and update hooks. Use it for hard calls, strategy, hiring, investments, major life choices, or when you want a premortem or red-team review.

How this skill works

When invoked, the skill classifies the decision (stakes, reversibility, horizon, competence) and selects a compact set of mental models (4–8) to apply. It runs a two-track analysis: a rational mechanics track (economics, trade-offs, expected value, second-order effects) and a psychological track (biases, incentives, execution risk). The output includes an outside view, model scan, premortem, recommendation with confidence, reversal conditions, and concrete next actions.

When to use it

  • You need an ‘oracle take’ or a hard verdict on a major choice
  • Preparing a shareable decision memo or board-quality recommendation
  • Running a premortem, postmortem, or calibrated forecast register
  • Stress-testing a plan, partnership, hire, product launch, or investment
  • Sanity-check: “What am I missing?” before committing resources

Best practices

  • Start by stating the decision, objective, time horizon, constraints, and top unknowns
  • Use only the smallest useful set of models (4–8) and explain why each matters
  • Always produce an outside view or state you lack one; prefer base rates to vibes
  • Separate process quality from outcome luck; give rough numbers where helpful
  • Invert before concluding and include a reversal clause: what would change your mind

Example use cases

  • Quick Take: fast verdict with confidence, three reasons, biggest risk, one missing fact, and next step
  • Oracle Review: outside view, inside view, model scan, bias & incentive audit, premortem, recommendation
  • Decision Memo: board-ready recommendation with assumptions, failure modes, and next actions
  • Premortem/Forecast Register: explicit failure scenarios, probability bands, update triggers, and kill criteria

FAQ

I use low/medium/high based on robustness, competence boundary, and sensitivity to missing evidence; I avoid precise percentages unless supported by data.

When will you ask clarifying questions?

If the decision is high-stakes or under-specified, I will ask up to five targeted questions before giving a firm recommendation; for speed, I will proceed with explicit assumptions.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
worldly-wisdom skill by openclaw/skills | VeilStrat