review

Launch parallel code, performance, and specification review agents and consolidate findings.

Updated Apr 3, 2026
One-click install
npx skills add https://github.com/YunosukeYoshino/harness --skill review-yunosukeyoshino
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/YunosukeYoshino/harness/tree/main/.claude/skills/review
Command: npx skills add https://github.com/YunosukeYoshino/harness --skill review-yunosukeyoshino

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill enables automated, parallel review of a repository focusing on code quality, performance, and specification consistency, delivering a unified set of actionable findings.

Core Features & Use Cases

  • Parallel triage: Launch code-reviewer, perf-reviewer, and spec-reviewer in parallel to maximize feedback speed.
  • Consolidation & gating: Deduplicate findings, sort by severity, and gate on Critical/High findings before proceeding.
  • Actionable reporting: Produce a structured review in work/reviews/active with clear remediations and owners.

Quick Start

Run the review skill to launch three reviewers in parallel and generate a consolidated report.

Frequently Asked Questions about review

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

FAQPage Schema
How do I automate code review for quality, performance, and specification alignment?

Parallel code review launches three agents simultaneously to assess code quality, performance, and specification alignment. It deduplicates overlapping findings and gates on Critical or High issues before returning a consolidated report.

What is the best way to consolidate code review findings from multiple reviewers?

Consolidating code review findings requires collecting outputs from parallel quality, performance, and specification agents, deduplicating overlaps, and sorting by severity. The process gates on Critical or High findings before returning a prioritized report.

Does automated code review gate on critical and high severity findings?

Automated code review gates on Critical and High severity findings before proceeding. The process sorts deduplicated findings by severity and blocks workflow progression if any Critical or High issues are detected.

How do I generate a prioritized code review report with clear remediation steps?

Generating a prioritized code review report requires running parallel quality, performance, and specification agents, deduplicating findings, and sorting by severity. The consolidated report is written to work/reviews/active with clear remediations and owners.

Can I review repository performance and code quality in parallel?

Reviewing repository performance and code quality in parallel is achieved by launching perf-reviewer and code-reviewer agents simultaneously alongside a spec-reviewer. This maximizes feedback speed and consolidates all findings into actionable remediations.

What are the limitations of parallel code review agents for repository evaluation?

Parallel code review agents focus exclusively on code quality, performance, and specification alignment. The process gates on Critical or High severity findings, halting workflow progression until these issues are remediated before returning a consolidated report.