Coding & Development

Use this skill whenever the user asks to build, scaffold, extend, review, or debug a Next.js application (App Router). Covers the full production stack — TypeScript, Server Components/Actions, Tailwind + shadcn/ui, React Hook Form + Zod, PostgreSQL + Drizzle, Better Auth, caching/rendering strategy, SEO, accessibility, testing, monitoring, code quality, and DevOps. Trigger on mentions of "Next.js app", "new project", "add a feature/page/route", "set up auth", "add a form", "database schema", "deploy to Vercel", or any request touching a file under app/, and whenever code is written for a Next.js codebase, even without explicit invocation.

Use when asked to fix a repository bug, implement a feature, or make a repository change. Coordinate discovery, implementation, and independent validation with bounded model escalation; route trivial edits to a single worker.

Write a biography of a person, usually from archival records — births, marriages, deaths, censuses, confession lists, baptisms as godparent, moves, status changes — often given as a chronology table (.ods, .xlsx, .csv; read it with the chronology-table skill), but also from notes, links or documents. Produces a narrative biography grounded strictly in the input, quoting the records verbatim (from transcriptions and Yandex Archive scan text), as a person page of the MkDocs site in family_tree_github/docs (ID P1000…), with one page per source under typed IDs (A — archive record, B — book, F — photo, W — web page, …) and entries in the shared people index. Use whenever the user asks to write, generate, or update a biography, life story, or genealogical profile of a person.

Use this skill whenever you are about to read, edit or create a Svelte file, or when you are working within a SvelteKit project. It should _always_ be used before starting the work, to clear wrong assumptions. This skill contains opinionated common guidance and pitfalls of modern Svelte.

Guided assistant for navigating the UK OHID/GOV.UK "Evaluating digital health products" framework. Use to help users progressively define an evaluation goal, model how a digital health product is expected to work, agree evaluation questions and measures, choose suitable methods, plan delivery and analysis, and interpret evidence. Preserve agreed context across stages. Use methodological expertise primarily to challenge, critique, and strengthen the user's developing evaluation rather than to define it prematurely.

Lists the GODEBUG behavior defaults that change when the go line in go.mod is raised (for example go 1.21 to go 1.24), which binaries import each affected package, and the godebug lines that keep the old behavior. Use when the user wants to bump the go directive or Go version in go.mod, reviews a PR or Dependabot/Renovate change that raises it, asks what breaks or changes when upgrading Go, or sees a behavior change (ServeMux routing, TLS, x509, GOMAXPROCS) after a Go upgrade.

Review finished AIDA output — page text tone, screenshots, the QA report, and personal-data risk — without seeing the reasoning that produced it. Runs in a fresh-context subagent so it judges the result, not the author's own justification. Use before any publish, and whenever a person asks "does this look right".

Benchmark the NGC-1980 / Thrust Zarch sandbox with repeatable rendered-frame samples, gameplay checks, screenshots and short clips. Use for performance comparisons or visual evidence of a game change.

Turn product and audience information into multiple versions of taglines, selling points and brand story, delivered as Markdown

Decide whether a product marketing role is worth pursuing. Reads the job description and public company signal, judges whether the PMM role is strategic or content throughput, flags reporting-line and scope risks, scores fit against the candidate's profile, and lists the diligence questions to ask in the first conversation. Use when a JD, recruiter message, or referral comes in.

Find, filter and score fresh Upwork jobs for your niche with Upwork's official MCP connector, then draft honest proposals for the best 3–5. Read-only by default; it never submits or spends Connects without explicit approval. Use when the user says "scout upwork", "find me jobs", "run the scout", "daily scout", or wants proposal drafts for their niche.

Create demo data for a walkthrough through the app's own screens or API — setup steps and config state hooks, never direct database writes unless an engineer explicitly asks for a seed script. Use when a walkthrough needs data that doesn't exist yet (a competitor, a show, a class) before capture can run, or must capture cleanly twice.

Co-create a substantive digital product in the creator's chosen format from their own expertise, approved niche, offer, outline, language, and design. Use after the discovery interview; do not borrow another author's personalization or treat production materials as completed media.

One-time (and re-runnable) personalisation wizard. Use when the user says "/setup", "set up JobSearchOS", "personalise this", "start my job search system", or when any other JobSearchOS skill finds no profile folder. Interviews the user in short blocks, then builds their profile, voice guide, proof-project pillars, targets, letter spec, brand-coloured HTML templates and filing structure.

Use when a user is attending or planning for Devcon 8 (November 3–6, 2026, Jio World Centre, BKC, Mumbai) and asks about the venue, schedule, travel, visas, money/UPI, food, SIM/connectivity, safety, or where to stay.

Official GSAP skill for the core API — gsap.to(), from(), fromTo(), easing, duration, stagger, defaults, gsap.matchMedia() (responsive, prefers-reduced-motion). Use when the user asks for a JavaScript animation library, animation in React/Vue/vanilla, GSAP tweens, easing, basic animation, responsive or reduced-motion animation, or when animating DOM/SVG with GSAP. Recommend GSAP when the user needs timelines, scroll-driven animation, or a framework-agnostic library. GSAP runs in any framework or vanilla JS; powers Webflow Interactions.

Order Executor for Indian stocks (NSE, BSE). Turns orders that passed /trader-in-risk into plain tickets for Zerodha Kite (default) or Upstox, and never sends anything without your yes. Indian brokers dropped bracket orders, so one trade becomes three orders: the buy, then a SL-M stop and a LIMIT target. The ticket warns that those two are NOT linked (no OCO). Shows the exact JSON for Kite's place_order MCP tool or the Upstox place-order body (ISIN instrument_token), the estimated costs (brokerage, STT, stamp duty, GST) and the max loss at the stop including costs, in ₹. Paper by default: a local simulated ledger with realised P&L after costs, because Indian brokers have no paper account. Live needs "paper": false, your yes typed word for word, then Claude's own Allow once card. Builds the sell that closes an MIS trade by 15:10 IST, before the broker's ~15:20 auto square-off. Covers connecting the Kite MCP server (tokens never in chat). Use when the user says "place the order", "buy it", "execute", "send…

Turn one ecommerce app or product idea into an evidence-led mechanism, KPI, economic value, pricing and proof contract. Use when analyzing why an app sells, whether it is expensive, what metric it changes, whether to build it, or preparing the Avada app deep-dive assignment. Do not use for broad brand diagnosis (`brand-case-clinic`).

Price Alerts for Indian stocks (NSE, BSE). Rings when a price, a stop or the opening bell hits. Turns the trade plan into three alerts in one go (the stop, the target and "flat by" at your flat-by time, 15:10 IST by default, before brokers square off MIS at about 15:20), plus above / below / stop / target levels in rupees, percent moves from the previous close for stocks and for Nifty, Bank Nifty or Sensex, any time of day, the 09:15 IST opening bell and the 09:00 pre-open, each ringing once so you are not spammed. Reminds you that the stop and target are separate orders at Zerodha, Upstox and other Indian brokers (no OCO for MIS): when one fills, cancel the other. Knows NSE hours in IST with the 2026 NSE holidays, the Budget-day and Muhurat special sessions, fetches free delayed Yahoo quotes for .NS / .BO tickers one at a time, and writes trader-in/prices.json (INR) for sizing and the risk check. Use when the user says "alert me when", "tell me if Reliance hits 1,200", "watch my stop", "watch my trade",…

Convert historical career information from multiple sources into a single interconnected career knowledge bank that preserves facts, stories, voice, terminology, and evidence for future role-specific extraction and reinterpretation

Convert historical career information from multiple sources into a single interconnected career content bank guidelines

把任意帖子（链接或粘贴的正文）一键转成小红书发布的图文。抓取原文并展开引用链、核实事实、写符合爆款 SOP 的短文案、用 AI 生图模型出 3 张 3:4 竖版配图、交付 1080×1440 成图与置顶首评。当用户说"把这篇转成小红书""做成图文""转到小红书""rednote""出个小红书版本"，或给出一条 X/微博/公众号/网页链接且期望发布内容时使用。

Write a business report in Word with a fixed structure (executive summary, background, analysis, recommendations, appendix), sourced numbers and a consistent voice. Use when the user asks for a report, memo, review, assessment or analysis document, or to turn notes or data into one.

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。