adversarial-review

Name: adversarial-review
Availability: InStock
Author: Bora

Identifies concrete failure scenarios in code changes via iterative refutation.

248|32|Updated Mar 30, 2026

Search Tags:#verification #code-review #adversarial #invariants #diff-analysis #failure-scenarios

Ready to run

npx skills add https://github.com/boraoztunc/skills --skill adversarial-review-boraoztunc

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: adversarial-review
Source: https://github.com/boraoztunc/skills/tree/main/adversarial-review
Command: npx skills add https://github.com/boraoztunc/skills --skill adversarial-review-boraoztunc

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Adversarial review helps teams evaluate code changes by assuming they are broken and forcefully attempting to refute them, surfacing concrete failure scenarios before shipping.

Core Features & Use Cases

Systematically generate failure scenarios that survive refutation passes.
Break down large diffs into focused domains (backend, frontend, tests, integration) for targeted checks.
Produce a structured, human- and machine-readable report to guide remediation and verification.

Quick Start

Run an adversarial review on the current diff to surface concrete failure scenarios that survive refutation.

Frequently Asked Questions about adversarial-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I find concrete failure scenarios in a pull request before merging?▼

Adversarial code review identifies failure scenarios by assuming a code change is broken and attempting to refute it across backend, frontend, and integration boundaries. It iteratively challenges findings until they survive scrutiny, producing a structured verdict set for remediation.

What is adversarial code review and how does it work with diff analysis?▼

Adversarial code review is a verification technique that evaluates diffs by forcefully attempting to refute their correctness. It breaks down large diffs into focused domains like backend, frontend, and tests, then iteratively challenges the code until only valid failure scenarios remain.

How do I enforce project invariants when reviewing a branch diff?▼

To enforce project invariants during a branch diff review, apply an adversarial review process that assumes the changes violate these constraints. The review iteratively refutes the diff against your invariants, generating a structured report of confirmed violations.

Can I use automated failure scenario generation for frontend and backend integration boundaries?▼

Yes, adversarial review covers frontend, backend, tests, and integration boundaries by breaking large diffs into focused domains. It systematically checks each area for failure scenarios that survive refutation passes.

What is the best way to structure code review findings for automated verification and reporting?▼

The best way to structure code review findings for reporting is to output a structured, machine-readable verdict set. Adversarial review produces this format after iteratively refuting failure scenarios, making the results suitable for direct synthesis and automated reporting.

Does adversarial review work on large diffs or is it limited to small pull requests?▼

Adversarial review works on large diffs by breaking them down into focused domains such as backend, frontend, tests, and integration boundaries. This targeted breakdown allows the refutation process to systematically handle extensive code changes across entire branches.