What is the Steel-Man Obligation feature?

It requires the AI to articulate the strongest possible counter-argument to its own conclusion to test if the verification survives adversarial scrutiny.

How does evidence tiering improve accuracy?

By classifying information into tiers like 'Verified' (primary sources) and 'Alleged' (single source), it prevents speculative or single-source information from being presented as absolute fact.

When will this skill NOT trigger?

It is designed to ignore requests for opinions, creative writing, how-to instructions, or when a user explicitly requests a quick, non-verified answer.

What is the 'Ralph-Wiggum' methodology in this context?

It refers to an iterative workflow philosophy that asks 'Is this actually verified, or did I just say it is?' and repeats investigative loops until genuine completion thresholds are met.

Can this skill help with conflicting data sources?

Yes, it uses a 'Dialectic Spiral' protocol to compare thesis and antithesis across multiple rounds until a stable, verified resolution is reached.

Iterative Factual Verification

Name: Iterative Factual Verification
Author: bogheorghiu

bybogheorghiu

0•

Aprendizaje y Documentación

Ensures factual accuracy through a rigorous multi-pass investigation framework and evidence-tiering methodology.

The Iterative Verification skill (utilizing the 'Ralph-Wiggum' methodology) transforms Claude from a single-pass search agent into a rigorous investigative researcher. It enforces a strict loop logic where claims are labeled into specific evidence tiers—Verified, Credible, Alleged, or Speculative—and subjected to adversarial search protocols. By requiring specific thresholds for source count, evidence freshness, and depth of investigation, this skill systematically eliminates confirmation bias and superficial reporting, making it indispensable for high-stakes analysis, technical auditing, and investigative research.

Características Principales

01Automated Threshold Enforcement (Sources, Freshness, Depth)

020 GitHub stars

03Evidence Tier Labeling (Verified, Credible, Alleged, Speculative)

04Probability Distribution Outcome Modeling

05Adversarial Search & Steel-Man Protocols

06Dialectic Spiral for Resolving Contested Claims

Casos de Uso

01Fact-checking complex current events and news reports

02Conducting deep-dive investigative research with multiple conflicting sources

03Auditing technical documentation or marketing claims for accuracy

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add bogheorghiu/ex-cog iterative-verification

For use in Claude.ai and ChatGPT

Características Principales

01Automated Threshold Enforcement (Sources, Freshness, Depth)

020 GitHub stars

03Evidence Tier Labeling (Verified, Credible, Alleged, Speculative)

04Probability Distribution Outcome Modeling

05Adversarial Search & Steel-Man Protocols

06Dialectic Spiral for Resolving Contested Claims

Casos de Uso

01Fact-checking complex current events and news reports

02Conducting deep-dive investigative research with multiple conflicting sources

03Auditing technical documentation or marketing claims for accuracy

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add bogheorghiu/ex-cog iterative-verification

For use in Claude.ai and ChatGPT