Why does the skill ask to quantify beliefs with percentages?

Quantification removes ambiguity. It is easier to resolve a 40% gap in probability than it is to resolve vague terms like 'unlikely' versus 'risky,' which mean different things to different people.

What are 'common priors' in this context?

Common priors are the initial assumptions or base-rate expectations agents hold about a situation before new, specific evidence is introduced.

How does this skill help in software development?

It provides a structured framework to resolve technical or strategic conflicts by determining if a disagreement is due to different information, different starting assumptions, or communication gaps.

What is Aumann's Agreement Theorem?

It is a mathematical theorem stating that rational agents with the same initial beliefs (priors) cannot 'agree to disagree' about a probability if their updated beliefs (posteriors) are common knowledge.

Can I use this for disagreements about personal values?

No, the theorem applies to factual beliefs and probability-based predictions. It is not intended for subjective preferences or value-based disagreements.

Aumann's Agreement Theorem

Name: Aumann's Agreement Theorem
Author: lev-os

bylev-os

0•

协作工具

Facilitates rational consensus by identifying and resolving the roots of persistent disagreements between team members using Bayesian logic.

This skill implements Aumann's Agreement Theorem as a diagnostic framework to help teams navigate complex disagreements. By assuming that rational agents with shared information should converge on similar conclusions, it guides users through quantifying beliefs as probabilities, identifying divergent priors, and uncovering hidden information asymmetries. It is particularly useful when stakeholders reach conflicting conclusions from the same data set, providing a structured path toward consensus or a clear diagnosis of why a disagreement persists, such as different starting assumptions or irrational biases.

主要功能

010 GitHub stars

02Diagnoses whether conflicts stem from differing priors, hidden data, or irrationality

03Identifies information asymmetries by systematically comparing underlying evidence

04Analyzes 'common knowledge' levels to ensure all parties share the same mental model

05Facilitates structured information exchange to encourage belief convergence

06Quantifies vague disagreements into specific Bayesian probability estimates

使用场景

01Resolving conflicting technical architecture recommendations between engineers

02Identifying hidden assumptions in high-stakes cross-functional decision making

03Aligning product and engineering teams on feature success probability

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lev-os/agents aumann-agreement-theorem

For use in Claude.ai and ChatGPT

Download Skill