review-adversarial

What problem does it solve?

This Skill exposes argumentative, evidentiary, and framing weaknesses in manuscripts by running multiple hostile-but-fair reviewer personas so authors can prioritize fixes before submission or publication.

Core Features & Use Cases

• Sequentially runs six adversarial personas that target rhetorical bias, logical fallacies, strawmaning, overcomplication, clinical relevance, and weakest-link claims.
• Resolves file globs, enforces guards for empty scopes, collects per-persona findings, and produces an aggregated convergence report highlighting cross-persona vulnerabilities and resilient claims.
• Use cases include pre-submission manuscript red-teaming, internal quality assurance for research chapters, and automated stress-testing of clinical recommendations.

Quick Start

Ask the skill to run an adversarial review on the manuscript glob typst/**/*.typ and produce a consolidated vulnerabilities report in your runtime directory.