reflexion:critique

Coordinate three judge agents to review work and produce a consensus report.

Updated Mar 4, 2026

Search Tags:#LLM #multi-agent #consensus #review #code-quality #debate #critique

Ready to run

npx skills add https://github.com/dalawwa/labor-methods --skill reflexion-critique

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: reflexion:critique
Source: https://github.com/dalawwa/labor-methods/tree/main/.cek/plugins/reflexion/skills/critique
Command: npx skills add https://github.com/dalawwa/labor-methods --skill reflexion-critique

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Enables teams to perform thorough, multi-perspective reviews of completed work by coordinating multiple specialized judges to independently critique, debate, and form a consensus on quality, correctness, and improvement opportunities without automatic fixes.

Core Features & Use Cases

Orchestrates independent judge agents (Requirements Validator, Solution Architect, Code Quality Reviewer) to assess alignment with requirements, design, and code quality.
Supports Phase-based workflow: Context gathering, parallel judging, cross-review debates, and consensus reporting.
Provides a structured consensus report outlining strengths, gaps, and actionable next steps for teams.

Quick Start

Instruct the system to critique the latest work or a specified set of files to generate a comprehensive, multi-perspective evaluation.

Frequently Asked Questions about reflexion:critique

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

FAQPage Schema

How do I run a multi-agent code review to evaluate requirements, architecture, and code quality?▼

To run a multi-agent code review, this Skill coordinates three specialized judge agents to independently assess requirements alignment, solution architecture, and code quality, then resolves differences through cross-judge debate to produce a final consensus report.

What is multi-agent critique and how does debate-driven consensus work for software deliverables?▼

Multi-agent critique is an evaluation method where independent judge agents review software deliverables in parallel, debate their findings, and synthesize a structured consensus report with explicit scoring and actionable recommendations.

Can I use this multi-agent review workflow for design proposals and code changes, or is it only for completed code?▼

Yes, you can apply this multi-agent review workflow to design proposals and code changes as well as completed software deliverables, preserving a read-only workflow that generates evaluations without applying automatic fixes.

Does the multi-agent consensus review automatically fix documented gaps in my code?▼

No, the multi-agent consensus review does not automatically fix documented gaps; it operates a read-only, review-only workflow that outputs explicit scoring, identified gaps, and actionable next steps for your team to address.

What is the best way to get a structured evaluation report with scoring and actionable recommendations for my project?▼

The best way to get a structured evaluation report is to instruct the system to critique your latest work, which triggers context gathering, parallel judging, cross-review debates, and consensus reporting with documented strengths and gaps.

Do I need to install any dependencies or components to use the multi-agent critiquing workflow?▼

No, you do not need to install any dependencies or components to use the multi-agent critiquing workflow, as it operates natively to coordinate the specialized judges and generate the consensus evaluation report.