best-of-n-with-judge

Generates, tests, reviews, and selects the best code variant via blind voting.

11|3|Updated Apr 23, 2026
One-click install
npx skills add https://github.com/BB-84C/omo-slim-superpowers-patch-kit --skill best-of-n-with-judge
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: best-of-n-with-judge
Source: https://github.com/BB-84C/omo-slim-superpowers-patch-kit/tree/main/opencode-config/skills/best-of-n-with-judge
Command: npx skills add https://github.com/BB-84C/omo-slim-superpowers-patch-kit --skill best-of-n-with-judge

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill streamlines the process of generating, vetting, and selecting the best implementation among multiple candidates by automating parallel code generation, testing, review, and consensus procedures.

Core Features & Use Cases

  • Parallel candidate generation: Creates multiple implementation options simultaneously, reducing manual effort.
  • Automated testing and validation: Runs tests and linting to gate candidates and ensure quality.
  • Blind review and voting: Uses structured oracle assessments to evaluate candidate approaches objectively.
  • Decision arbitration: Resolves ties via council arbitration or repeats with feedback.
  • Use Case: Ideal for high-stakes code refactoring where multiple design options are explored, and the best solution must be objectively determined.

Quick Start

Command the AI to generate multiple solutions for a specified task, then automatically test, review, and select the best candidate, handling tie-breaks and cleanups seamlessly.

Frequently Asked Questions about best-of-n-with-judge

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

FAQPage Schema
How do I automate code review for multiple parallel implementations?

Automated code review for parallel implementations uses structured oracle assessments to objectively evaluate multiple candidates. It generates options simultaneously and applies automated testing and linting to gate candidates for high-quality outputs.

What is the best way to select the best code variant among multiple candidates?

Selecting the best code variant involves automated judging and voting procedures that evaluate multiple candidates. It uses decision arbitration via council arbitration or repeats with feedback to resolve ties and ensure the highest quality solution.

Can I use automated testing and linting to gate parallel code generation candidates?

Yes, automated testing and linting gate parallel code generation candidates by validating quality before review. This ensures only candidates passing test and linting checks proceed to blind review and voting.

How does decision arbitration work when voting results in a tie?

Decision arbitration resolves voting ties via council arbitration or by repeating the generation and review cycle with feedback. This structured approach ensures a single, objectively determined solution is selected.

When do I need parallel experimentation for high-stakes code refactoring?

Parallel experimentation for high-stakes code refactoring is needed when multiple design options must be explored objectively. It enables simultaneous candidate generation and automated evaluation to efficiently determine the best solution.

Does parallel code generation work with git worktrees for isolated testing?

Parallel code generation leverages git worktrees to enable isolated testing and automated validation of multiple variants. This allows structured review and voting processes to evaluate each candidate independently without conflicts.