judgment-day

Runs blind dual-judge adversarial code review with bounded fix and re-judgment rounds.

Updated Mar 7, 2026
One-click install
npx skills add https://github.com/albersg/dotfiles --skill judgment-day-albersg
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: judgment-day
Source: https://github.com/albersg/dotfiles/tree/main/dotfiles-opencode/.opencode/skills/judgment-day
Command: npx skills add https://github.com/albersg/dotfiles --skill judgment-day-albersg

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Single-reviewer code reviews can miss defects or produce unverified findings. This Skill orchestrates an explicit dual review where two blind, read-only judges independently inspect the same immutable target, so only findings confirmed by both judges trigger fixes. ## Core Features & Use Cases - Blind Dual Review: Launches two independent judges in parallel against an identical immutable target with identical criteria, then merges their findings into a frozen ledger. - Bounded Fix Rounds: Applies corrections only for severe findings confirmed by both judges, with at most two fix rounds and two scoped re-judgments. - Escalation Discipline: Contradictions between judges or issues surviving round two escalate to a human decision instead of looping indefinitely. - Use Case: Before merging a critical pull request, request Judgment Day to get an adversarial double review that ends in a single terminal verdict: JUDGMENT: APPROVED or JUDGMENT: ESCALATED. ## Quick Start Ask the assistant to run Judgment Day dual review on the specific files or changes you want verified.

Frequently Asked Questions about judgment-day

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

FAQPage Schema
How do I run a dual adversarial code review?▼

Explicitly request Judgment Day for a concrete target. The orchestrator freezes the target, launches two blind read-only judges in parallel with identical scope and criteria, merges their findings into a frozen ledger, and only fixes severe issues both judges confirm.

What happens when the two review judges disagree?▼

A finding reported by only one judge is recorded as a suspect and is never auto-fixed. When judges directly contradict each other, the review escalates for an explicit human decision rather than resolving the conflict automatically.

How many fix rounds does Judgment Day allow?▼

At most two fix rounds and two scoped re-judgments are permitted. Re-judgment sees only the frozen ledger plus the fix delta, and any severe issue remaining after round two forces an escalated terminal state with no third round.

When should I use Judgment Day instead of a normal review?▼

Use it when a target needs adversarial corroboration, since two-judge agreement replaces the ordinary review flow for that target. It is not appropriate for quick informal checks, because it requires explicit activation and a full dual-judge transaction.

What verdicts can a Judgment Day review produce?▼

The only terminal states are approved or escalated, emitted as exactly one final line: JUDGMENT: APPROVED or JUDGMENT: ESCALATED. The output also includes confirmed, suspect, contradiction, and info counts plus artifact references.