Recently Added
Fresh skills just added to the collection. Be the first to try them out!
semver-release-recommender
by Nick2bad4u
Analyze repository changes from the last release to the current revision and recommend the correct semantic version bump. Use when the user asks what version to release, whether a release should be patch/minor/major, or wants release-impact analysis before tagging or publishing.
platform-skills
by nitinjain999
"Use when troubleshooting, implementing, reviewing, or auditing platform infrastructure as a system — where Kubernetes, GitOps, CI/CD, and security concerns intersect. Provides structured diagnosis with blast radius, validation steps, and rollback plan for: Kubernetes, Flux CD, Argo CD, Terraform, GitHub Actions (composite actions, OIDC, SHA pinning), AWS, Azure, GKE, Linkerd, KEDA, Karpenter, supply chain security (Cosign, SBOM, SLSA), Falco, Chaos Engineering, DORA metrics, Datadog/Dynatrace/LLM observability, SOC 2, and PR review."
antigravity-lazy-packs
by albertchang1008-alt
AntiGravity 懶人包 — 服務連接與工作流程設定。說「AntiGravity 懶人包」「安裝 Antigravity 設定」時載入。
universal-data-acquisition-pipeline
by Pranjay-kumar
Trigger when the user wants to collect, structure, evaluate, crawl, extract, refresh, or build reusable data acquisition pipelines from websites, APIs, portals, files, or rendered apps. Use for dataset design, source classification, feasibility, endpoint discovery, authorized/owned-session scraping plans, Playwright fallback, source probing, pagination analysis, scraper/pipeline architecture, sample validation, refresh design, and output contracts. Do not trigger for ordinary browsing, exploitative access, credential theft, CAPTCHA solving, auth bypass, rate-limit bypass, or non-data tasks.
atmospheric-science-research
by zsy1207
Use for climatology data processing, diagnostic computation, and publication-grade plotting.
exampass
by MIKUZ12
将课程资料（PPT/Word/PDF/截图）按章节生成深度知识清单、教授式详细讲解和交互式章节测试，帮助学生高效期末复习。
wind-mcp
by jwangkun
万得 Wind 金融数据 MCP 技能 — 44 个工具覆盖 A 股/港美股/基金/指数/债券/公告/宏观/分析 8 大服务域
claw-hancomdocs
by DoHyun468
Drive 한컴독스(Hancom Docs) web viewer/editor (webhwp) via Playwright to upload .hwp/.hwpx files, capture rendered pages (A4 screenshots / zoomed regions / find-by-text), and directly edit documents through Hancom's own web UI. Use when the user wants to SEE how a Korean document renders in 한컴독스, screenshot a hwp/hwpx page, "한컴독스에 올려서 보여줘", check styling/tables/layout as the web viewer shows them, detect that a file won't open, or create/edit a document through Hancom's native web editor. Headless, reuses saved login. Local-machine only — does not run in Cowork sandbox (proxy blocks webhwp.hancomdocs.com).
appendix3a-quiz-publisher
by babusarkar
"Generates, reviews, and publishes Railways Accounts Appendix 3A exam MCQ quiz questions to the @appendix3a Telegram channel. Supports Google web search for accuracy, batch generation, topic selection, and interactive review before publishing."
pharos-agent-escrow
by zaikaman
Use when an AI agent needs to create, fund, accept, complete, verify, release, refund, or audit an on-chain work order escrow on Pharos. Supports agent-to-agent task payments with native PHRS/PROS or ERC20 assets, work metadata, proof submission, event-based reputation, and explorer links.
Funder Report Builder
by SNEHA-Org
Never write to the database; connect read-only.
aicoin-trading
by aicoincom
"CEX 中心化交易所(Binance / OKX / Bybit / Bitget 等)的下单交易工具。严格规则:(1) 所有订单必须通过 node scripts/exchange.mjs create_order 执行,禁止写自定义代码下单 (2) create_order 分两步:第一次返回预览,展示给用户等确认,用户说确认后第二次加 confirmed=true 执行 (3) 禁止自动确认,禁止跳过预览 (4) 平仓必须用 close_position,禁止用 create_order 构建平仓单。Trigger 关键词: 'buy on okx', 'sell on binance', '在 OKX 买 BTC', '在 Binance 下单', '做多 BTC 永续', '杠杆做空 ETH', '平掉我的 SOL 仓位', 'CEX 下单', '现货买入', '合约开仓', '永续平仓', '止盈止损', 'long', 'short', 'leverage', '买', '卖', '下单', '做多', '做空', '开仓', '平仓', '平掉', '关仓'. 路由提示: 用户说“链上 swap / Uniswap / DEX 买 PEPE / Solana 上买”是链上 DEX 交易,应走 aicoin-onchain 而不是本 skill. Hyperliquid 上的下单也走 aicoin-onchain(HL 是链上 perp DEX),不是这里. 本 skill 只处理 CEX 现货 + 永续合约下单。"
rally-brand
by Arnarsson
Apply RALLY's brand identity to any artifact — app screens, slides, posters, social posts, store assets, emails. RALLY is an editorial "football-culture meets fashion" brand: warm paper/light theme, heavy grotesque headlines, an italic editorial-serif accent, and a lime + pink/blue/purple pop palette with jersey/streetwear motifs. Use whenever designing or styling anything for RALLY, or when the user mentions RALLY brand, brand colors, brand feel, or wants something "on brand".
sentinel
by RayLi-Git
一個貫穿全程的工程師思考 OS。用 76 個世界名校思考習慣 + 11 個自創資安習慣（共 87 個：批判 29／創意 17／溝通 10／互動 20／資安 11）當作 Claude Code 寫程式時的底層思考指令集，從「規劃→開發→診斷→解決→復盤」五階段全程把關，專治 vibe coding 最大的病：在淺層打轉、貼補丁繞過警報、改了症狀沒動根因，以及把金鑰寫死、不驗證輸入、給過大權限等資安僥倖。任何「寫 code、規劃專案、改架構、修 bug、測試一直失敗、轉換率/效能下降、處理使用者資料/金鑰/認證、想找盲點、要做技術決策」的工程任務都應觸發此 skill。它會在動手前喊停做淺層 vs 深層自檢與資安紅旗檢查、按任務份量三級調節思考力道、自動把夠痛的除錯經驗寫進專案病歷（.claude/debug-log.md + patterns.md）並產出根因樹，讓 Claude Code 越用越不踩同一個坑。關鍵詞：sentinel、根因、淺層打轉、vibe coding、debug、資安、注入、機密管理、最小權限、思考習慣、根因樹。
seo-audit
by Cosmofang
"Audit and optimize any website's SEO + GEO (AI/LLM visibility), then fix what's broken. Runs a portable, zero-dependency hard-gate audit over a build output directory (Astro/Next/Hugo/Jekyll/plain HTML) and a live-URL crawl/GEO audit (robots.txt AI-crawler policy, sitemap, llms.txt, on-page JSON-LD, canonical, security headers). Use when asked to: check/improve a site's SEO, raise a Lighthouse/PageSpeed SEO score, make a site discoverable by AI agents (ChatGPT/Claude/Perplexity/Gemini), add structured data, set up robots/sitemap/llms.txt, or review a site before launch. Distilled from the reference site's production build-time SEO gates."
token-cost-optimization
by protick-bjit2019
Use when you need to reduce LLM API token usage or costs — prompt compression, model tiering, caching, context trimming, tool call minimization, output length control, and automated context compression via headroom-ai.
tmux-mysql
by lenovobenben
Use when the current prepared tmux pane is already inside an interactive mysql-compatible client REPL (mysql>, mariadb>, MySQL [db]>, MariaDB [db]>, or a configured mysql prompt), including MySQL, MariaDB, TiDB, Percona Server for MySQL, StarRocks, OceanBase MySQL mode, ShardingSphere-Proxy, or similar backends operated through the mysql client. Safely send read-only SQL and explicitly user-approved insert/update/delete SQL without using Linux shell wrappers, inspect prompt state, cancel broken multiline input, and exit the client REPL.
react-design-draft
by EdwardWason
"Generate React design drafts (4-piece set) from content. Invoke for 'design draft'/'设计稿'/'生成页面'/'信息图'/'知识卡片'/'多图配图'. Do NOT use for editing existing code."
vue-js-lowcode-conversion
by ljh-0206
'Vue<->低代码JS转换：模板、styleClass、_this.info、commonsJs 语义还原'
site-art-direction
by Ianbrogan2
Art direction and anti-"AI-slop" design law for building ianbrogan.com (Ian Brogan's personal site). Use this skill whenever designing, building, styling, or refining ANY page, section, component, or visual element of this website. It enforces a distinctive "Markets & Vineyards" editorial aesthetic, photographic vineyard realism, the use of the owner's real photos, reference-matching, and a hard ban on generic AI defaults.
kirbs-design
by dawooni0625
KIRBS를 위한 잘 브랜딩된 인터페이스와 에셋을 생성할 때 이 스킬을 사용하세요 — 깊은 인디고(#2D3378) 프라이머리, Pretendard 한글 서체, 완전한 W3C-DTCG 토큰 레이어를 갖춘 B2B 어드민/콘솔 디자인 시스템입니다. 프로덕션 코드, 일회성 프로토타입, 목업, 슬라이드 덱, 그리고 KIRBS다움이 필요한 디자인 탐색에 적합합니다.
pitchy
by rdmgator12
"Adversarial business-plan / pitch-deck / strategic-memo reviewer for founder work. Use this skill ONLY when the user explicitly invokes Pitchy or asks for adversarial / rip-it-up review. Trigger phrases (explicit only — do NOT fire on generic 'review my plan' or 'critique my deck'): 'pitchy', 'pitchy audit', 'pitchy review', 'pitchy round 1', 'pitchy round 2', 'pitchy round 3', 'pitchy full audit', 'pitchy multi-agent', 'rip up this plan', 'rip it up', 'adversarial review', 'adversarial audit', 'run pitchy on X', 'hit it with pitchy'. Pitchy is structurally adversarial — finds load-bearing claims that can't survive contact with reality. Output: defect table (P1/P2/P3 severity + lane), top fatal flaws, the one thing to change first, verdict (PASS / PASS WITH LIMITATIONS / FAIL—UNSUPPORTED / FAIL—INTERNALLY INCONSISTENT / FAIL—UNSAFE OVERCLAIM), and v-next actionables. Two modes: (1) single-agent rounds 1→2→3 for iteration on v0.1 drafts; (2) multi-agent full audit (orchestrator + 3-5 parallel specialists: finance, market, strategy, optional compliance + product) for deck/board/investor prep. Pitchy is the strategy-artifact lane: not med-legal review, not clinical reasoning, not operational planning, not technical code/architecture review, not validation-seeking. v0.5.0 adds Pattern 1 — author-administered Step 0 prohibited; methodology author must delegate Step 0 claim-currency preflight to an independent administrator when reviewing their own thesis."
owner
by benditobr
Owner da Bendito. Acione esta skill quando o usuário disser "owner", "decisão semanal", "fechar a semana", "o que fazer essa semana", "rodar o owner", "prioridades da semana", "decisão estratégica", "síntese dos shapers", "questionar o modelo", "revisar tese", "foco atual", "atualizar foco", "reunião de projeto", "o dono decide", "visão estratégica", "challenger", "me questiona", "revisão de direção", "onde estamos", "o que cortar". Este agente consolida inputs dos Shapers (Finance, Operation, Brand) e de Projetos específicos, questiona o que precisa ser questionado e entrega decisão — não sugestão.
barescript
by craigahobbs
Teaches a model how to write BareScript — language syntax, the built-in library, the include library, MarkdownUp client-rendered applications, and unit tests. Use this skill whenever the user asks you to write, read, modify, or review BareScript (.bare) code, a MarkdownUp Markdown application, a markdown-script code block, or any code that uses BareScript built-in functions (array*, object*, string*, data*, markdownPrint, elementModelRender, drawNew, etc.). Also use it when asked to write unit tests for BareScript code. Apply this skill even if the user does not say the word "BareScript" — recognize it from the .bare extension, the markdown-script fenced code block, or the function-naming style above.