How does it calibrate severity ratings?

It uses a standardized matrix based on Impact and Likelihood, ensuring that critical issues like data loss or security breaches are flagged consistently regardless of the individual reviewer.

How does this skill handle duplicate findings from different reviewers?

It applies specific merge rules: if multiple reviewers find the same issue at the same location, it consolidates them into one finding while crediting all reviewers and retaining the most detailed description.

Can I use this for PR summaries?

Yes, the skill includes a professional Markdown template for generating consolidated reports that are ready to be posted as pull request comments or documentation.

What review dimensions are supported by default?

The skill includes built-in focus areas for Security, Performance, Architecture, Testing, and Accessibility, with logic to recommend specific combinations based on the file type.

Multi-Reviewer Patterns

Name: Multi-Reviewer Patterns
Author: wshobson

bywshobson

•

29,555

•

コラボレーションツール

Orchestrates comprehensive code reviews by coordinating multiple quality dimensions and consolidating findings into actionable reports.

The Multi-Reviewer Patterns skill empowers Claude to act as a sophisticated code review orchestrator, managing parallel evaluations across security, performance, architecture, and testing. It solves the complexity of multi-agent reviews by implementing strict finding deduplication rules, consistent severity calibration, and unified reporting templates. This skill is ideal for teams looking to automate high-stakes code audits while ensuring a coherent, non-redundant feedback loop for developers.

主な機能

0129,555 GitHub stars

02Context-aware dimension recommendations based on the type of code changes

03Standardized severity calibration for consistent risk assessment

04Consolidated reporting with cross-referenced locations and reviewer attribution

05Multi-dimensional review allocation across security, performance, and architecture

06Automated finding deduplication to prevent redundant feedback items

ユースケース

01Performing deep-dive security and performance audits for critical API endpoint changes

02Standardizing code quality benchmarks and severity ratings across large engineering teams

03Consolidating feedback from multiple AI reviewers into a single, clean pull request report

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add wshobson/agents multi-reviewer-patterns

For use in Claude.ai and ChatGPT

Download Skill