skills.rest
roderik
roderikCOMMUNITY
Explore all skills by this author →
Website𝕏@roderikGitHub

adversarial-review

Spawn opposite-model reviewers via CLI to critique code changes and output a synthesized verdict.

Updated Feb 24, 2021
Search Tags:#code-review#adversarial-review#opposite-model#reviewer-lenses#lens-based-review
One-click install
npx skills add https://github.com/roderik/roderik.github.io --skill adversarial-review-roderik
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: adversarial-review
Source: https://github.com/roderik/roderik.github.io/tree/main/.agents/skills/adversarial-review
Command: npx skills add https://github.com/roderik/roderik.github.io --skill adversarial-review-roderik

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Adversarial-review provides a structured, multi-lens critique of code changes by spawning reviewers on the opposite model to challenge the work, ensuring potential flaws are surfaced before shipping.

Core Features & Use Cases

  • Spawns 1–3 reviewers on the opposite model (Claude vs Codex) to challenge changes from distinct critical lenses.
  • Produces a synthesized verdict summarizing findings without making changes.
  • Enforces hard constraint: reviewers must run via the opposite model's CLI (codex exec or claude -p) to ensure genuine cross-model critique.
  • Uses three reviewer lenses (Architect, Skeptic, Minimalist) and linked references to ground judgments.
  • Ideal for large diffs, multi-file changes, or planning sessions requiring thorough, multi-perspective evaluation.

Quick Start

Initiate an adversarial review of the latest changes and generate a verdict.

Frequently Asked Questions about adversarial-review

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

FAQPage Schema
How do I perform adversarial code review on large multi-file diffs?▼

Adversarial code review on large diffs works by spawning 1–3 reviewers on the opposite model to critique code changes from distinct critical lenses. It then synthesizes findings into a verdict with structured findings and recommendations without modifying the code.

What is cross-model code review and how does it find flaws?▼

Cross-model code critique works by running reviewers via the opposite model's CLI, ensuring genuine independent evaluation. It applies Architect, Skeptic, and Minimalist lenses to surface potential flaws before shipping.

Can I use lens-based code review for planning sessions?▼

Lens-based code review applies to planning sessions requiring thorough, multi-perspective evaluation. It spawns reviewers on the opposite model to challenge the proposed changes and outputs a synthesized verdict.

Do I need the opposite model's CLI to run multi-lens code review?▼

Multi-lens code review enforces a hard constraint that reviewers must run via the opposite model's CLI, such as codex exec or claude -p. This ensures genuine cross-model critique rather than same-model evaluation.

What's the best way to critique code changes before shipping?▼

The best way to critique code changes before shipping is an adversarial review approach that spawns multiple reviewers on the opposite model. This multi-lens evaluation produces a synthesized verdict with structured findings and concrete recommendations.

Related Skills

Explore More Software Engineering Skills→
haryelramalhoharyelramalho

adversarial-review

Spawn opposing-model reviewers to critique code changes and plans.

Community
Advanced
bricerisingbricerising
7

review

Run a four-phase adversarial code review with evidence-backed findings and verdicts.

Community
Advanced
mengxi-reammengxi-ream
9.0k

adversarial-review

Spawn opposite-model reviewers to critique code diffs and synthesize verdicts.

Community
Advanced
rodrigobranasrodrigobranas
1

adversarial-review

Spawn opposing-model reviewers to critique code diffs and synthesize verdicts.

Community
Advanced