skills.rest
ahuaracab
ahuaracabCOMMUNITY
Explore all skills by this author →
GitHub

judgment-day

Runs blind dual-agent adversarial code review with fix and re-judgment loops.

Updated Aug 31, 2026
Search Tags:#code review#pull request#multi-agent#quality gate#adversarial review#ai judges#dual review
One-click install
npx skills add https://github.com/ahuaracab/test-genesis --skill judgment-day-ahuaracab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: judgment-day
Source: https://github.com/ahuaracab/test-genesis/tree/main/.agents/skills/judgment-day
Command: npx skills add https://github.com/ahuaracab/test-genesis --skill judgment-day-ahuaracab

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Single-reviewer code reviews miss defects and suffer from individual bias. This Skill orchestrates two independent judge agents that review the same target in parallel without seeing each other's verdicts, then synthesizes only confirmed findings before any fix is applied. ## Core Features & Use Cases - Blind Dual Review: Launches two judges in parallel with identical prompts and criteria, then classifies findings as confirmed, suspect, contradiction, or INFO. - Fix and Re-Judge Loop: Delegates confirmed fixes to a separate fix agent and immediately re-runs both judges to verify no regressions, ending only in APPROVED or ESCALATED states. - Warning Triage: Distinguishes real warnings triggerable by normal use from theoretical ones, downgrading the latter to INFO. - Use Case: Before merging a large PR touching shared test fixtures, trigger Judgment Day to get an adversarial dual verdict, fix confirmed criticals, and re-judge until approved. ## Quick Start Ask the AI to run Judgment Day on the current diff or a specific set of files before creating the pull request.

Frequently Asked Questions about judgment-day

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

FAQPage Schema
How do I run an adversarial dual code review on a pull request?▼

Trigger Judgment Day explicitly by naming the target files, diff, or PR. It launches two blind judge agents in parallel with identical criteria, synthesizes their findings, and only treats issues both judges found as confirmed.

What is blind dual review in AI-assisted code review?▼

Blind dual review means two independent judge agents evaluate the same code without seeing each other's verdicts. This prevents cross-contamination, so agreement between judges signals a real issue rather than copied reasoning.

When should I not use dual-agent adversarial review?▼

Skip it for trivial typo, formatting-only, or autogenerated diffs, since orchestrating two judges plus re-judgment costs more than the verdict is worth. Also split diffs larger than 30 files into focused per-area reviews.

How does the review handle warnings that are hard to trigger?▼

Warnings are classified as real only if normal intended use can trigger them. Contrived or impossible-path warnings are downgraded to INFO as theoretical, and theoretical-only findings do not block approval.

What happens after the judges confirm issues in the code?▼

The Skill asks before fixing Round 1 confirmed issues, delegates fixes to a separate surgical fix agent, then immediately re-launches both judges. After two fix iterations with remaining issues, it asks whether to continue or escalate.

Related Skills

Explore More Software Engineering Skills→
chriswchchriswch
2

clarifying-intent

Elicit intent and constraints from underspecified product requests into structured specs.

Community
Advanced
FourthWizFourthWiz
2

end_of_task

Finalize approved development tasks by committing changes, pushing branches, and archiving artifacts.

Community
Advanced
arndvsarndvs

pr-preflight

Audits changed files, diffs, and stack tooling before opening a pull request.

Community
Advanced
p-o-ke-naep-o-ke-nae

quick-fix

Apply small, targeted fixes to an existing pull request branch.

Community
Advanced