takaoumehara

superforge-biz

Architect business models, pricing tiers, paywall placement, subscription flows, and Go-To-Market plans. Covers monetization archetype selection, price anchoring, trial-to-paid conversion triggers, and growth loops. Use when the user says "pricing", "monetization", "paywall", "subscription", "free trial", "business model", "revenue", "GTM", "how do we make money", "価格", "値付け", "課金", "マネタイズ", "サブスク", "ペイウォール", "ビジネスモデル", "どう収益化する", or runs /superforge-biz.

superforge-brain

Run the BreakBias engine — an exhaustive, machine-checkable idea sweep. Decompose the problem across five lenses, name the hidden bias on every element, ban the obvious three, then push every element through eight transformation techniques and their sub-methods as a tracked cell ledger, so no combination is silently skipped. Survivors are killed by explicit codes, judged in a separate context, and only then checked against the market. Use when the user says "brainstorm", "ideas", "come up with", "what could we build", "reframe this", "concept", "something more original", "アイデア", "発想", "ブレスト", "企画", "コンセプト", "ありきたりじゃないもの", "何が作れる", "BreakBias", "SIT", "虱潰しで考えて", or runs /superforge-brain.

superforge-brand

Define brand identity systems (naming, tone, palette, type, logo direction) and produce AI-generated visual and motion assets through prompt engineering for generate_image, Kie.ai, and Higgsfield. Use when the user says "brand", "logo", "identity", "tone of voice", "hero image", "banner", "generate an image", "video concept", "moodboard", "ブランド", "ロゴ", "トンマナ", "世界観", "ビジュアル", "画像を作って", "動画", or runs /superforge-brand.

Code & Development Solid

superforge-debug

Root-cause-first debugging with FailForward learning memory, which recalls past failures before diagnosing and records new ones after. Forbids trial-and-error patching and dummy fallbacks that mask symptoms. Use when the user says "bug", "error", "it crashes", "build fails", "test is failing", "not working", "why does this happen", "still broken", "バグ", "エラー", "動かない", "落ちる", "直らない", "なぜか失敗する", or runs /superforge-debug.

superforge-dev

Build multi-component features by decomposing the work and dispatching subagents, assigning a model tier per subtask across Claude 5, Gemini 3.6, Codex, and Kimi. Proposes Subagents versus Agent Teams topology by task complexity and token cost. Use when the user says "implement", "build this feature", "execute the plan", "in parallel", "dispatch agents", "subagents", "which model should", "実装して", "作って", "並列で", "サブエージェント", "プランを実行", "どのモデルで", or runs /superforge-dev.

superforge-roast

Deliver an unsparing, compliment-free critique of a design, PRD, plan, architecture, copy, or UI to expose flaws before they ship. Strips AI politeness and names the weakest point first. Use when the user says "roast", "be brutal", "tear this apart", "what's wrong with", "honest feedback", "critique", "poke holes", "don't be nice", "辛口で", "ダメ出し", "批評して", "欠点を指摘", "忖度なしで", or runs /superforge-roast.

Web & Frontend Solid

superforge-ui

Design and build interfaces across Web, iOS (SwiftUI), and Android (Jetpack Compose), covering layout, visual hierarchy, typography, responsive behaviour, state specification, and micro-interactions. Follows a five-phase design process, Apple HIG, and Material 3. Use when the user says "design", "UI", "UX", "layout", "screen", "component", "make it look better", "animation", "spacing", "typography", "デザイン", "画面", "見た目", "レイアウト", "アニメーション", "余白", "使いにくい", or runs /superforge-ui.

superforge

Concierge and router for the whole making process, from idea to shipped product. Reads intent, hands the work to the right superforge-* skill (brain, biz, brand, ui, dev, test, debug, roast, verify, handoff), and assigns a model tier per subtask across Claude 5, Gemini 3.6, Codex, and Kimi before any agent is dispatched. Use at the start of any build, or when the request spans several of those areas. Use when the user says "let's build", "I want to make", "help me ship", "where do I start", "何か作りたい", "作って", "どこから始める", "一気に進めたい", or runs /superforge.

superforge-handoff

Write the session state to a structured capsule in .handoff/ so any AI model or tool (Claude Code, Codex, Gemini CLI, Antigravity, Cursor) can resume the work with zero context loss. Use before clearing a long thread or switching tools. Use when the user says "handoff", "save the session", "switch models", "before I clear", "pick this up later", "context is getting long", "引き継ぎ", "ハンドオフ", "セッションを保存", "モデルを切り替える", "コンテキストが長い", or runs /superforge-handoff or /handoff.

Testing & QA Solid

superforge-test

Drive implementation through Red-Green-Refactor TDD across Web (Jest, Vitest, Playwright), iOS (Swift Testing, XCTest), and Android (JUnit, Compose Test), verifying each red and green state by actually running the suite. Use when the user says "write tests", "TDD", "test first", "add test coverage", "refactor safely", "テストを書いて", "テスト先に", "TDD", "リファクタリング", "テストが無い", or runs /superforge-test.

superforge-verify

Gate that blocks any completion claim until there is empirical runtime evidence — the test suite actually run, both mobile and desktop viewports checked, and native iOS or Android simulator runs confirmed. Use before saying a task is done, fixed, or shipped. Use when the user says "is it done", "did it work", "verify", "double check", "are you sure", "before we ship", "確認して", "検証して", "本当に動く", "終わった", "動作確認", or runs /superforge-verify.