Coding & Development

老馬的全能剪輯助手（Laoma Omni Editor）。把真人口播原片，或一篇文字稿（AI 旁白），自動做成成片：轉錄、剪輯、字幕、依內容挑圖解與動態效果、配樂、品檢、渲染；支援 16:9 長影音與 9:16 短影音（直式）。使用者說「老馬的全能剪輯助手」「做口播影片」「幫我剪這支口播」「做短影音」「做長影音」「把這篇稿做成旁白影片」並提供口播影片或稿件時使用。給不熟 Claude 的同事用：你全程代勞，只在幾個關卡停下來問簡單問題。

Reference-coding with XERJ. Clone the libraries that already solved your problem, index them locally, and retrieve the exact implementation before writing code — so the agent reads passages instead of re-deriving algorithms across retry loops. Use when starting a task in an unfamiliar API, porting an algorithm, or when you have already looped twice on the same error.

程序化短片——把一个主题变成一部竖屏动画短片：画面全部用 JavaScript 在 canvas 上绘制，音乐与音效都由 Web Audio 合成，并在结束时生成可直接发布的封面图、标题与描述。当用户要求制作一部程序化短片、一部关于某个主题的手绘风格短动画，或需要"一个主题 → 一条可直接发布的竖屏视频"时使用。

Find empty niches in any domain by building a quantitative concept map of ~50-150 real items (products, apps, games, services, tools, papers...). Items are coded on 8-15 explicit dimensions with a written rubric and a blind re-coding check; the map comes from Ward clustering, PCA gradients and UMAP (visualization only); gaps are found as absent value combinations and low-density zones, filtered by the target user's profile, translated into concrete ideas and validated against the market with a mirage test. Use it when someone asks "what is missing in this market/genre?", "where is there room for a new product?", "map the landscape of X and show me the gaps", or wants product ideas backed by a verifiable corpus. Do NOT use it to size a market in money, to profile a single competitor, or to extract jobs-to-be-done from interviews.

Design system and anti-patterns for professional UI. Ensures apps don't look "AI generated". Defines color palettes, typography, spacing, animations, and common anti-patterns to avoid. Used by Design Reviewer sub-agent to polish UI. Triggers: design review, polish UI, improve aesthetics, looks like AI made it, design system, style guide, make it beautiful.

Render structured interactive UI inline through the dsh-ui fence. Use for key points, emphasis, comparisons, flows, steps, status, data, demos, and interactions whenever structured presentation would be clearer than prose. Preserve conversation language for all user-visible text.

Build or migrate an app marketing website from a supplied prototype using Bun, following the AceCut static website convention, and prepare or complete Vercel hosting and an app subdomain. Use for app official websites; not native app development.

Second step, after ai-product-start. Interviews the product manager about the value the product gives customers at each of six levels of AI, from System of Record to System of Understanding, maps the market on the same levels, sees who's moving in the market, maps the work around the product and ranks the opportunities in and around it, then interviews the product manager one question at a time to write the product's AI vision. Writes ai-levels.md, market-map.md, market-map.html, market-moves.md, opportunities.md and vision.md in the product folder.

Your launch lead. Builds a dated, step-by-step launch plan: a cheap real-world test first (a waitlist, a pre-sale or a pop-up with a real price, so real buyers confirm what the simulated panel said), then the countdown to opening day with owners and deadlines, the launch-day run sheet, and the first 30 days of what to measure and what to change. Use when the user says "how do I launch", "launch plan", "when can I open", "launch checklist", "go live", or "soft open".

Audit and directly improve HTML or WordPress pages for technical SEO, on-page SEO, accessibility-related discoverability, structured data, internal linking, and generative engine optimization (GEO) or AI-search readability. Use when Codex must implement evidence-based search improvements, metadata guidance or integration, schema/JSON-LD, semantic content structure, and citation-friendly answers without keyword stuffing, fabricated facts, or guaranteed rankings.

Writes paid ad copy for Meta (Facebook and Instagram), Google Search, and LinkedIn: several variants, each testing a different hook, sized to each platform's limits. Use whenever someone wants ads, ad headlines, PPC or paid social copy, or variants to A/B test, even for one quick ad.

Generate a realistic live challenge for a company and role (design challenge, product case, take-home, system design, campaign brief) and critique the user's approach against how interviewers score it. Use when the user says "/challenge-prep [company]", "I have an interview at", "give me a practice challenge", "whiteboard prep", "take-home prep", "case prep", or when status first moves to screening. Auto-triggered by /log-application at that transition when the module is on.

Turn positioning and launch context into sales-ready assets: product playbook, one-pagers, talk tracks, FAQs, objection handling, demo scripts, and training plans, scaled to the launch tier. Use when packaging enablement for sales, CS, or partners, or when a founder needs to arm their first sellers.

Build or restyle PowerPoint decks (.pptx) in Pomelo Digital's house style — white canvas, Georgia headlines, Calibri body, coral + mint accent system, corner arc motif, and the pómelo wordmark. Use for any Pomelo client report, performance report, pitch deck, or proposal, and whenever asked to "build/redo/restyle a deck in Pomelo style." Provides a reusable PptxGenJS helper module (pomelo.js), the logo asset, and a runnable starter example.

Cria e evolui sistemas com login, banco de dados, painel ou API própria no template guibean/nextjs-fullstack-template. Usa Next.js 16 standalone, PostgreSQL, Drizzle, Better Auth, Zod, Base UI e deploy em VPS.

A skill that localizes contents into given locales.

MiniMax Music 3（2026-08公開のオープンな曲生成モデル）で、歌詞と曲の説明から歌入りの曲（最大約5分、日本語の歌詞も可）をローカルGPUで作る。「曲を作って」「MV用のオリジナル曲」「この歌詞で歌にして」「BGMを作曲して」などの依頼で使う。できた曲はbiim-motionスキルのMVモードにそのまま渡せる。

Optimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.

Turn a topic into a fact-checked 2–3 minute knowledge film whose pictures are drawn by code you write. For any city's life story the `model` style flies from the globe into the real city in 3D (every OpenStreetMap building at its height) and plays its history on it as light, era by era, with narration-synced effects (rivers, walls, fires, links, rebuilding). Ink-scroll and Roman-atlas packs serve people, landmarks and animals. Only a text LLM (you, the Agent) and tools are used; facts are cross-checked across publishers; narration and a music track through SandBase are opt-in.

Plan and score a product or feature launch: pick a launch tier, build a phased plan (T-60 to T+30), check readiness across messaging, enablement, support and measurement, and flag what's missing before go/no-go. Use for launch plans, GTM readiness reviews, release tiering, or launch checklists.

Use whenever working with EffectCraft on this machine — the Rust motion-graphics / VFX compositor and After Effects alternative — including launching the app with control port ("EffectCraftを起動", "EffectCraftを立ち上げて", "launch EffectCraft"), creating or editing compositions, layers, keyframes, masks, tracking, rotoscoping, effects and text (.ecproj, MP4/MOV/WebM/ProRes/EXR/PNG sequences, .ectemplate, .ecpreset, ExtendScript/.jsx), driving the running desktop window, or using its MCP server, effectcraft-cli or control channel. Triggers on "EffectCraft", "effectcraft-cli", "effectcraft mcp", "EffectCraft MCP", "EffectCraftを起動", "EffectCraft立ち上げて", "EffectCraft開いて", "make a motion graphic", "animate this text", "add an effect", "track this shot", "rotoscope", "export to MP4/ProRes/EXR", "take an EffectCraft screenshot", and requests to automate motion graphics, VFX or EffectCraft itself. Invoke with @effectcraft.

Build a Stripe-style hero wave in Three.js r178: stripe.com's own folded sheet (a plane folded in two, thick with a round edge), on the settings stripe.com ships, with Stripe's unlit shading (a palette painted along the sheet, long silky streaks, a pale underside), big and cropped by its frame, every part drifting on its own noise. Use whenever the user asks for a 3D hero visual, a "Stripe wave", a shader ribbon, a WebGL model or shape, or wants hero art for an idea or brand. Input is the meaning, never a shape: ask for what the product does (one sentence), the brand colours, and where the art sits; with no prompt, read the colours from the code and use Stripe's light-hero wave. The idea changes the wave's colour, crop and settings, never its form.

Coaches a content marketer to analyze a piece of B2B content the way a senior content reviewer would. Walks the learner through a Socratic review (gut reaction → a batch of priority questions → one consolidated senior-reviewer take) and lands on a clear path forward (keep, promote, edit, restructure, rewrite, kill, or repurpose), so the learner leaves with an action, not just an analysis. Use whenever someone wants to check their critique of a draft or published piece, decide whether to fix or kill an underperforming piece, or says things like "what would a senior reviewer think of this", "review my blog", "tear this apart", "help me critique this use case page", "this piece has low traffic, is it worth fixing", or "help me get better at analyzing content". Use for blog articles, use case pages, customer success stories and interview articles. Not for social posts or content briefs.

Quality-check translated software strings (English → Spanish by default) before they ship. Runs a script that catches broken placeholders ({{name}}, %1$s, %@, ICU plurals), lost HTML or XLIFF tags, changed URLs and numbers, strings over a character limit, missing or extra keys, untranslated strings, missing ¿ ¡, then reviews what a script can't see — register (tú/usted), terminology consistency, meaning, gender and plural agreement around placeholders, and locale formats. Use whenever someone shares localization files (JSON, iOS .strings or .xcstrings, Android strings.xml, gettext .po, XLIFF/.xlf) or asks to "check", "QA", "review", "lint" or "validate" a translation of an app, website, game or UI. Also trigger for "did I break any placeholders", "will this fit the button", "revisar la localización", "QA de strings". Not for translating from scratch (use translator-en-es) or for RAE rule disputes in prose (use rae-spanish).