Build crafted, highly detailed, interactive isometric SVG figures that explain a technical idea as a real physical object (a test rig, a dial gauge, a cabinet of drawers, a plotter, a lock), with many small precise solid parts, four-tone shading, correct painter's order, a framed card with title, hint and live readout, and calm eased motion. Use this whenever someone wants an isometric illustration, an explanatory diagram or figure for docs, a blog post, a landing page or an article, a "Linear-style" or "premium" technical drawing, an interactive explainer, or an SVG that shows how a system, algorithm or component works, even if they never say "isometric". Works framework-free (TypeScript/JS that writes SVG and HTML) and in React/Next.js, and can add realistic WebGL shaders (fire, exhaust, water, caustics, lamp light) that live inside the drawing in the same isometric perspective. Also publishes a finished figure to the Anatomy community gallery when asked.
Researches a question on the web, cross-checks at least two sources and reports findings with citations. Use when the user asks to look up, search, compare or verify facts, prices, news or documentation (查詢、搜尋、比較、調查).
Converts books and documents (PDF, EPUB, DOCX, HTML, Markdown, plain text, RTF, MOBI/AZW with Calibre) into structured agent skills, extracting frameworks, mental models, principles, techniques, and anti-patterns. Use when the user wants to study a document through GitHub Copilot CLI, Amp, Claude Code, Hermes Agent, OpenCode, or OpenClaw, apply an author's frameworks while working, or build a reusable knowledge base from a file.
Analyses whether a technology shown in a video or described in text can work - extracts every testable claim, checks it against physics with a numeric model (required vs limit), gives a verdict (feasible, feasible within limits, infeasible) and explains the engineering and the physical limits. Then builds an interactive 3D / isometric anatomy page with the anatomy skill - how the system works and where it stops if feasible, or why each key property cannot be achieved if not - plus a Markdown report. Use for 技術可行性、工程原理、物理極限、這可能嗎、真的做得到嗎、科技產品或發表會影片分析, feasibility study, is this physically possible, how does it work, debunk or explain a tech demo.
Manages the agent's layered memory - distills the key points of a task into task, short-term, long-term or permanent memories, compresses memory when it grows, and adds or removes memories on the user's instruction. Use when the user asks to remember, forget, pin, review, clean up or compress memories (記住、永久記住、忘記、刪除記憶、整理記憶、壓縮記憶、你記得什麼), or at the end of a long task to keep what matters.
Watches a YouTube video (metadata, full transcript with timestamps, and stills about every 10 seconds) and writes a Markdown report with images - a SinoBee.ai-style cover, what the video says, a timestamped outline, an analysis section chosen from 16 video categories, and a fact-check of its claims for news, politics, science, finance, health, tech and documentary videos. Use when the user gives a YouTube link and asks to summarise, analyse, review, fact-check or report on it (YouTube 影片分析、影片摘要、幫我看這部影片、查核影片內容).
Turns a one-line idea into a finished AI video plus a reusable production kit (zip with script, characters, locations, style, music, voices, shot language and reference images), and makes style-consistent sequels or remakes from such a kit (film, documentary, vlog, commercial, music or explainer) with the Universal AI Video Pipeline - script and audio-first timing, subject profiling matrix, consistency assets, shot grammar and keyframes, camera and subject dynamics, then post-production - rendered with MiniMax H3 on a ComfyUI server and assembled with ffmpeg. Use when the user asks to make, shoot, generate or edit a video, short film, clip, trailer, ad, documentary or vlog (做影片、拍短片、生成影片、短影音、紀錄片、廣告片、預告片、MV).
Operates native desktop applications reliably with screenshots, mouse and keyboard. Use when a task involves a GUI app, window, dialog or anything the browser tool cannot reach (桌面、視窗、點擊、應用程式).
Logs in, fills and submits web forms and completes multi-step website flows with the browser tool, confirming before irreversible submissions. Use for sign-in, registration, booking, ordering or any web form (登入、表單、報名、預約、下單).
Explains a source-code repository (GitHub URL or local folder) - what it is for, architecture, tech stack, how to run it and its current state - grounded in files actually read. Use when the user asks to look into, explain, summarize or review a repo, project or codebase (專案、repo、程式碼、解說、介紹這個專案).
Writes and improves Agent Skills (SKILL.md) that follow the agentskills.io specification and SinoBee conventions. Use when the user asks to learn, remember or save a procedure as a skill, or after finishing a reusable non-trivial task (學會、記住這個流程、做成技能).
Guided modernization of a legacy codebase in any language - assessment, structure map, business rules mined into Given/When/Then cards with file:line citations, an approved phased plan, then an upgrade (uplift), rewrite (transform) or rebuild (reimagine) with script-checked proof that the new code behaves like the old, plus security hardening and a one-page report. Use for 舊系統現代化、系統翻新、改寫、升級框架或語言版本、COBOL/Java 8/.NET Framework/PHP 5/Python 2 遷移、業務規則萃取, legacy modernization, migrate, rewrite, upgrade, uplift.
Answers questions from the user's databases (PostgreSQL, MySQL/MariaDB, SQLite) - inspect the schema, write read-only SQL, show the SQL, then report tables and charts. Use when the user asks about data, numbers, customers, orders, trends or reports stored in a database (資料庫、查詢、報表、統計、SQL).
Turn a finished dimensional drawing sheet into a ~10 s Apple-style motion film — lines draw themselves in, the camera pulls back, layers lift into an exploded view, the full sheet resolves. Use after the dimensional-drawing skill, or when asked to animate a spec sheet or anatomy drawing.
Draws a large engineering plant (wastewater or water treatment, power, incineration, chemical or fab utilities) from one model as a zoomable SVG with a switchable isometric 3D view and top-down plan - modular facility groups built from their real structures, colour- and line-coded pipes per fluid with flow arrows, reusable defs components, names and process figures. Use for 廠區配置圖、污水處理廠、淨水場、管線流向圖、工程圖、鳥瞰圖、立體圖、俯視圖, plant layout, process flow layout, isometric plant diagram.
Turn one line ("a turbofan jet engine", "a mechanical watch", "an LVAD heart pump") into a photoreal, physics based, interactive 3D simulation of that machine in a single HTML file, plus a 20 to 30 second film to post it. Real proportions in metres, real materials (PBR metals, paints, polymers, glass), studio lighting with soft shadows, moving parts driven by a small solver that computes real numbers (thrust, torque, rpm, flow, pressure, efficiency), and modes to explode it, cut it open with clean section caps, x-ray it, slow it down, label it and break it. Use this whenever someone asks to build, simulate, visualise, explain or "show how X works" for any machine, engine, mechanism, device, instrument or piece of hardware, wants an interactive 3D model or a cutaway, or wants a video of a machine for a post, even if they never say "simulation" or "three.js".
Make an Apple-style monochrome dimensional drawing of a UI component or physical product — title-block summary plus one dense drawing sheet with variable-based dimensions. Use when asked for an anatomy, spec sheet, redline or dimensional drawing of a component.
Builds an immersive scroll-scrubbed "fly through the world" landing page for any industry or brand - as the visitor scrolls, a pre-rendered camera flies from outside each scene into its interior and on to the next with no cuts. Interviews the user (subject, story sections, brand kit, art direction, camera style, mobile, budget), generates cohesive scene stills and frame-locked camera clips with Higgsfield or Monid (Seedance 2.0), and wires a framework-agnostic vanilla-JS scroll-scrub engine. Use when the user wants a scrollable 3D world / diorama landing page, a scroll cinematic, or to turn a business into a fly-through hero (滾動式 3D 世界、捲動飛越、等角模型著陸頁、滾動電影感首頁).
把中文內容改寫成臺灣讀者習慣的繁體中文:簡體轉正體、大陸用語換成臺灣用語、標點依教育部《重訂標點符號手冊》、語氣改得專業自然(去掉空泛與「AI 味」的寫法),並逐條查證文中的事實、數字與引述,查不到的不當成事實寫。Use when the user asks to convert, localize, polish or proofread Chinese text for Taiwan (改成繁體、臺灣用語、在地化、潤稿、校稿、去 AI 味、查證內容、更專業).
技能怎麼運作
SKILL.md 寫明何時使用與步驟;scripts/、references/、assets/ 放腳本、參考與範例。上傳 SKILL.md 或 sinobee skills export 產生的 .tar.gz。
上傳時自動掃描危險指令;審核通過的版本以 ed25519 簽章,版本發布後不可修改。SinoBee 依 builtin、verified、community 等等級決定能否執行腳本。
請求中出現技能的觸發詞時,SinoBee 會直接推薦對應的技能與快速上手步驟,模型不用自己猜。
複雜技能可以宣告 YAML 流程(例如影片製作:企劃 → 草稿 → 核准 → 正式渲染),在「工作」頁一鍵執行、可排程、可續跑。
在 SinoBee 裡直接使用
SinoBee 內建 Hub 用戶端:搜尋、安裝、檢查更新與發布技能,並自動驗證內容雜湊、簽章與撤銷清單。用 export 打包含腳本與參考文件的技能,再 publish 上傳。
sinobee skills search video sinobee skills install hub:cafeca/ai-video-production sinobee skills outdated && sinobee skills update sinobee skills validate ./my-skill sinobee skills export my-skill # .tar.gz with scripts & references sinobee skills publish my-skill