Where are the review reports stored?

Reports are automatically saved to the repository at docs/pr-reviews/{branch-name}-{date}.md for easy access and version control.

How does it handle findings from multiple AI agents?

It includes specific deduplication rules to merge overlapping issues and uses agent-specific prefixes (like SEC- or ARCH-) to track which agent identified each finding.

Can this skill automatically block a pull request?

While it doesn't physically lock the git provider, it provides a Verdict Decision Matrix that outputs a clear 'Blocked' or 'Changes Requested' status based on issue severity.

What format does the PR review report use?

The skill generates a comprehensive, professionally formatted Markdown report including an executive summary, risk level table, and categorized issue lists.

How are issues prioritized in the report?

Issues are mapped into Critical, High, Medium, and Low severities, with Critical and High issues requiring detailed descriptions and code solutions while others are summarized in tables.

PR Review Output Formatter

Name: PR Review Output Formatter
Author: mrstroz

bymrstroz

•

コラボレーションツール

Standardizes and aggregates multi-agent pull request review findings into a unified, actionable markdown report.

This skill provides a standardized reporting framework for Claude Code to consolidate feedback from multiple specialized review agents—such as security, architecture, and performance scanners—into a single source of truth. It implements a rigorous decision matrix to determine PR verdicts, deduplicates overlapping findings from different agents, and organizes issues by severity to help developers prioritize fixes. By automating the creation of structured reports with clear remediation steps and audit trails, it streamlines the code review workflow and ensures consistent quality standards across a development team.

主な機能

01Severity-based issue categorization with agent-specific ID prefixes

021 GitHub stars

03Deduplication logic for overlapping findings from multiple agents

04Actionable remediation checklists for mandatory and recommended fixes

05Standardized Markdown report structure with executive summaries

06Automated verdict decision matrix for blocking or approving PRs

ユースケース

01Maintaining a historical audit trail of PR reviews within a project's documentation

02Consolidating feedback from multiple AI review agents into a single developer-friendly document

03Automating the PR approval process based on predefined risk levels and issue counts

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mrstroz/claude-code-plugins pr-review-output

For use in Claude.ai and ChatGPT

Download Skill