multi-agent-review

Orchestrate parallel specialist reviews of code diffs and pull requests.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/zcawood-mie/agentz --skill multi-agent-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: multi-agent-review
Source: https://github.com/zcawood-mie/agentz/tree/main/skills/multi-agent-review
Command: npx skills add https://github.com/zcawood-mie/agentz --skill multi-agent-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coordinating thorough, multi-perspective code reviews is time-consuming and inconsistent when performed by a single reviewer; this Skill ensures parallel, domain-specific coverage and a single, deduplicated outcome that surfaces the most important issues first.

Core Features & Use Cases

  • Parallel specialist dispatch: send the same diff and file context to Security, Architecture, Performance, and Maintainability reviewers concurrently.
  • Structured reviewer contracts: each reviewer receives a scoped prompt and a required return format so findings are consistent and machine-mergeable.
  • Merge, deduplicate, and prioritize: combine overlapping findings, tag which reviewers raised each issue, and sort by severity for clear triage.
  • Modes of operation: supports self-review (pre-push fixes), colleague-assisted reviews (draft comments), and PR-based reviews (draft GitHub reviews for approval).
  • Selective reviewer selection: choose a subset of specialists based on change type (e.g., Security + Architecture for auth/API changes).

Quick Start

Run the multi-agent-review skill on a pull request to produce a merged, severity-sorted review across security, architecture, performance, and maintainability.

Frequently Asked Questions about multi-agent-review

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

FAQPage Schema
How do I run a parallel code review across security, architecture, and performance domains on a GitHub pull request?

A parallel code review dispatches the same diff and file context to domain-specific reviewers concurrently, then merges, deduplicates, and severity-sorts findings into a unified review for GitHub PRs.

Can I choose specific reviewers for a code review instead of running all specialist assessments?

Yes, selective reviewer selection lets you dispatch a subset of specialists based on change type, such as running only Security and Architecture reviewers for authentication or API modifications.

What is the best way to deduplicate overlapping findings from multiple code reviewers?

Deduplicating overlapping findings is handled by aggregating specialist reviews, tagging which reviewers raised each issue, and sorting them by severity for clear triage in a single unified output.

Does multi-agent code review work with local diffs or only GitHub pull requests?

Multi-agent code review applies to both GitHub pull requests and local diffs, collecting full file context to support self-reviews, colleague-assisted reviews, and comprehensive cross-domain assessments.

How do structured reviewer contracts improve automated code review consistency?

Structured reviewer contracts provide each specialist a scoped prompt and a required return format, ensuring findings are consistent and machine-mergeable for deduplication and prioritization.

When should I not use a multi-agent approach for code review?

A multi-agent code review approach may be unnecessary for trivial changes or single-domain fixes where a comprehensive cross-domain assessment adds overhead without surfacing critical issues.