Magi review

Skill weiting-tw/MAGI-System/skills/magi-review

MAGI System — Inspired by Evangelion's MAGI supercomputer. Three AI engines (Claude, Gemini, Codex) analyze tasks in parallel, then reach consensus through voting. A Claude Code skills plugin.

Install

npx -y skills add weiting-tw/MAGI-System --skill magi-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

三機(Claude/Gemini/Codex)平行審查代碼，評估邏輯、品質、安全漏洞後投票共識

SKILL.md

1.7 KB, 678 tokens by cl100k_base, as published. Nobody here has run it

MAGI Review 快捷指令

/magi-review [code/PR] - 代碼品質審查

此指令為 /magi review [code/PR] 的快捷方式。

使用方式

/magi-review PR #123
/magi-review src/auth/login.ts

執行內容

自動啟動 MAGI 三機平行分析：

機體	引擎	分析重點
MELCHIOR-1	Claude Sonnet	邏輯正確性、架構一致性
BALTHASAR-2	Gemini Flash	代碼品質、命名一致性、錯誤處理
CASPER-3	Codex/GPT	OWASP Top 10、注入漏洞、機密洩漏

🚨 強制執行規則

<MANDATORY> 執行此指令時，必須嚴格遵守以下規則：

禁止事項：

❌ 禁止委派給其他 agent 處理
❌ 禁止使用非指定的工具

必須使用：

✅ MELCHIOR-1: Task tool with subagent_type="general-purpose", model="sonnet"
✅ BALTHASAR-2: mcp__gemini__ask-gemini (先用 ToolSearch 載入)
✅ CASPER-3: mcp__codex-cli__ask-codex (先用 ToolSearch 載入)

執行步驟：

先以單一 ToolSearch 載入 MCP 工具：ToolSearch(query="select:mcp__gemini__ask-gemini,mcp__codex-cli__ask-codex")
在單一訊息中同時發出三個 tool calls（Task + Gemini + Codex）
等待三機回應後彙整共識報告 </MANDATORY>

模型設定（review 任務）

機體	模型	參數
MELCHIOR-1	`sonnet`	-
BALTHASAR-2	`gemini-3-flash-preview`	-
CASPER-3	`gpt-5.2-codex`	`reasoningEffort: "medium"`

完整的投票機制與共識報告格式，見主 skill magi（skills/SKILL.md）。

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.