What is the difference between MAGI Quick and the full MAGI system?

MAGI Quick uses a single specialized agent for rapid feedback on straightforward, low-stakes items, while the full MAGI system employs a three-agent panel for comprehensive deliberation on complex topics.

When should I escalate a topic to full MAGI?

The skill will explicitly recommend escalation if the single agent determines the topic has enough complexity to warrant a multi-perspective 3-agent deliberation.

How does the skill choose which MAGI persona to use?

It selects an agent based on the topic: Melchior-1 for correctness and security, Balthasar-2 for maintenance and team impact, and Caspar-3 for strategy and feasibility.

Does it keep a record of the triage results?

Yes, the skill automatically persists a structured JSON log of the deliberation in the .magi/history directory for auditability and later review.

MAGI Quick Triage

Name: MAGI Quick Triage
Author: ohta-rh

byohta-rh

•

生产力与工作流

Provides rapid, single-agent technical assessments for low-stakes decisions using specialized MAGI personas.

MAGI Quick Triage is a streamlined decision-making framework designed to provide immediate feedback on technical proposals without the overhead of a full multi-agent deliberation. By dynamically selecting between the Melchior (correctness), Balthasar (maintenance), or Caspar (strategy) personas, it delivers a focused critique, standardized scoring, and a clear verdict. This skill is ideal for developers needing a quick sanity check on code changes, architecture tweaks, or project directions, ensuring that only truly complex issues are escalated to more resource-intensive review processes.

主要功能

01Intelligent escalation logic that suggests full MAGI reviews for complex topics

02Persona-driven assessment based on specialized MAGI logic (Melchior, Balthasar, or Caspar)

03Rapid single-agent triage for efficient feedback on low-stakes technical decisions

043 GitHub stars

05Standardized output format including key findings, scores, and clear verdicts

06Automated structured logging of all deliberations to local JSON history files

使用场景

01Validating design feasibility before moving a proposal into implementation

02Evaluating minor code refactoring strategies for potential maintenance impact

03Quickly assessing security and performance implications of new library dependencies

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ohta-rh/magi-system magi-quick

For use in Claude.ai and ChatGPT