review-loop

Automate parallel GitHub code reviews with 0-10 scoring and issue triage.

Updated Apr 17, 2026

Search Tags:#automation #parallel #review #quality-assurance #code-review #issue-tracking

Ready to run

npx skills add https://github.com/ebal5/agent-skills --skill review-loop-ebal5

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: review-loop
Source: https://github.com/ebal5/agent-skills/tree/main/skills/review-loop
Command: npx skills add https://github.com/ebal5/agent-skills --skill review-loop-ebal5

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

複数視点の fresh-context reviewer を並列起動し、スコア 0-10 で triage、スコア閾値に応じて「対応」または「Issue 起票」を行い、新規 5+ 指摘が出なくなるまでループするレビュープロセスを自動化します。

Core Features & Use Cases

複数視点の並列レビュープロセスを起動して、diff や変更点を横断的に評価します。
スコアリング（0-10）に基づく triage と、閾値に応じた対応方針の決定を自動化します。
指摘が新たに 5 件以上出なくなるまで、ループと再評価を継続します。
チームのコード品質向上や外部レビュアーの視点を取り入れた品質保証の標準化に適用できます。

Quick Start

対象リポジトリの差分を与え、4視点の並列レビューループを開始して新規の高優先指摘がなくなるまで回します。

Frequently Asked Questions about review-loop

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

FAQPage Schema

How do I automate parallel code reviews to triage issues on GitHub pull requests?▼

Parallel code review automation launches multiple reviewer perspectives simultaneously to evaluate diffs and triage issues with a 0-10 score. It routes results to immediate fixes, new GitHub issue creation, or skip decisions based on defined score thresholds.

What is the best way to standardize quality assurance across multiple review perspectives?▼

Standardizing quality assurance across multiple review perspectives involves running fresh-context reviewers in parallel, scoring findings 0-10, and looping until fewer than five new high-priority issues appear. This ensures consistent, comprehensive diff evaluation.

Do I need a multi-agent reasoning framework to run automated issue tracking loops?▼

Yes, automated issue tracking loops require a multi-agent reasoning framework like Sonnet to evaluate diffs from multiple perspectives. The framework powers the parallel scoring, triage, and routing policy for immediate fixes or new GitHub issue creation.

How does scoring work when triaging code review findings from different perspectives?▼

Triage scoring in code review findings uses a 0-10 scale assessed by parallel reviewers. A defined threshold determines the routing policy: high scores trigger immediate fixes, moderate scores create new GitHub issues, and low scores are skipped.

When should I stop the review loop for automated issue tracking?▼

You should stop the review loop when parallel reviewers produce fewer than five new high-priority findings during an evaluation cycle. This threshold ensures the code diff has been thoroughly checked across all perspectives.