Recently Added
Fresh skills just added to the collection. Be the first to try them out!
yb-marketing-design
by kyledean-yb
Use this skill to generate well-branded interfaces and assets for YB Marketing (Yakima Branding), either for production or throwaway prototypes/mocks/etc. Contains essential design guidelines, colors, type, fonts, assets, and UI kit components for prototyping.
video-spec-builder-remotion
by mrzhangguoguo
当用户说想做一个视频、宣传片、产品演示、动画短片、抖音/YouTube 内容，或者说要改分镜、调节奏、换镜头、调字幕、加配音、改转场时使用。通过苏格拉底式追问收集视频需求，主动激发渲染层的全部能力（TTS / 字幕 / 3D / shader / 音频反应等），输出标准化的 video-spec.md 用于 Remotion 渲染。
agent-onboarding
by kaijie0074-art
新手上手 Agent 的交互式教程引擎。当用户想第一次体验 Agent 能干什么、需要被一步步带着完成一个真实任务、或说出"带我上手 / agent 上手 / 新手教程 / 教我用 agent / 我是新手不会用 / 体验一下 agent / 第一次用 / 手把手教我 / 我想做个微信群整理(文章转视频/竞品监控/经营日报/个人官网/自动作业/竞赛材料 等)但不会"时触发。本 skill 让 Agent 化身"上手教练"，用选项卡带用户选项目→自检环境(缺工具自动装)→先用样例跑通→再用真实数据→一步步动手做完→可推飞书→做完讲清"你只说了X、Agent做了Y、下次你只要这句话"。面向零基础、尤其语音交互用户。不要用于：已经熟练的用户直接下指令做事(那就直接做，不用进教程)。
heitang-kb-forge-skill
by HeiTang-HuaMei
Agent knowledge supply-chain Skill for producing standardized, auditable, retrievable knowledge packages.
worldcup-betting-analyst
by sahuan14
世界杯及赛前热身赛的"竞彩足球"诚实分析助手。当用户想分析世界杯/热身赛比赛、聊买球/足彩/竞彩、问某场比赛怎么买、胜平负/让球/比分/总进球/半全场/串关怎么选、或想知道未来几天有哪些值得关注的比赛时,务必使用本 skill。即使用户只是随口提到某两支球队的世界杯比赛、问"这场买谁/怎么串"、或说"帮我看看世界杯足彩",也应触发。核心原则:以诚实分析为本,绝不承诺稳赚,把足彩定位成花闲钱的娱乐。仅覆盖中国竞彩足球玩法。
mem-sync Skill
by Mombrane
Push rejected: run mem-sync sync first to pull latest, resolve any conflicts, then mem-sync flush again.
minecraft-modpack-server
by NousResearch
"Host modded Minecraft servers (CurseForge, Modrinth)."
pokemon-player
by NousResearch
"Play Pokemon via headless emulator + RAM reads."
baoyu-article-illustrator
by NousResearch
"Article illustrations: type × style × palette consistency."
pixel-art
by NousResearch
"Pixel art w/ era palettes (NES, Game Boy, PICO-8)."
baoyu-comic
by NousResearch
"Knowledge comics (知识漫画): educational, biography, tutorial."
hermes-s6-container-supervision
by NousResearch
Modify, debug, or extend the s6-overlay supervision tree inside the Hermes Agent Docker image — adding new services, debugging profile gateways, understanding the Architecture B main-program pattern.
pixel-design-taste
by shalhevetvardi
פיקסל — מעצבת UI/UX שנותנת טעימה חכמה מעיצוב מקצועי. שתי דלתות: (א) בדיקת "עיצוב AI" — מדביקים עיצוב/קוד/צילום מסך והיא מאבחנת דפוסים גנריים, נותנת ניקוד ותיקונים קונקרטיים; (ב) מיני-מסע לכיוון עיצובי — 3 שאלות מלמדות שמסתיימות בכרטיס בסיס ויזואלי אנטי-גנרי (פלטה, טיפוגרפיה, מרווחים, רגש). הפעל את הסקיל בכל פעם שמשתמש אומר "העיצוב שלי נראה כמו AI", "למה זה נראה גנרי", "תבדקי לי את העיצוב", "תני לי כיוון עיצובי", "פלטת צבעים", "אנטי גנרי", "פיקסל", "טעימה מעדי", או מדביק עיצוב/קוד/צילום מסך ומבקש פידבק עיצובי. גם באנגלית: "my design looks like AI", "anti-generic design check", "give me a design direction", "review my UI", "why does this look generic".
youtube-to-wechat
by abangzhu
把任意来源的内容变成一篇可发布的微信公众号文章，覆盖从字幕提取、原文整理、人物背景、内容分析、中文写作、自审重写到标题和封面图的完整流程。支持 YouTube 视频、网页文章、PDF/Markdown/TXT 本地文档。只要用户提供内容来源（URL、视频链接、文件路径、文本片段），并且目标是中文文章、公众号稿、翻译整理、长文改写、内容总结、视频/文章/PDF 转写，就使用此 skill。
jtbd-segmentation
by ivanintouch-labs
Универсальная методология JTBD-сегментации СВОИХ клиентов по СВОИМ разговорам (звонки/чаты/мессенджеры/боты). Self-serve для предпринимателя или продакта, который работает со своими данными — без discovery с внешним заказчиком. Извлекает из каждого диалога полное JTBD-досье на лида (per-lead JSON) и собирает 5-12 JTBD-сегментов с отнесением каждого лида к одному из них. Адаптируется к любому уровню глубины данных — от «только разговоры» до «полная воронка + конверсия». Используй, когда нужно сегментировать своих клиентов по реальным мотивам (не по демографии и не по CRM-типу), отнести каждого лида к сегменту с описанием и evidence. Триггеры — «сегментировать клиентов», «JTBD из звонков», «JTBD из переписок», «понять кто наши клиенты», «отнести лидов к сегментам».
decision
by beyond06120925
人生决策顾问 — 基于61个科学框架帮你做决策。触发词：我要不要、怎么选择、帮我分析
letsgo
by Clarkky1
This skill should be used when the user types "/letsgo", says "letsgo", "let's go build", "start a new project", or "kick off a new build". It runs the full 15-step project kickoff flow: personal vs client check (with client meeting prep if client), gather context + MCP setup, brand check + design system, UI standard (Hallmark/portfolio-design/motion-design-school/website-ui-v2/liquid-glass-design/landonorris-ui/pixel-art) + design-inspiration brief (Mobbin/Godly/Awwwards/Lapa/VibeUI/Aceternity/Origin UI/Magic UI/Motion Primitives/21st.dev/Fancy Components/hover.dev/buildui.com/UIverse/Emil Kowalski/Grainient/Mesher/CSS Pattern/SVG Backgrounds/Lummi/Fontjoy/Typewolf/Realtime Colors/oklch/Radix Colors/Lucide/Phosphor/Tabler/60fps.design/Codrops/transition.style), platform detection + animation stack + lenis + matter.js, 3-terminal setup (with iteration cap: 4 attempts max per sub-agent then escalate to supervisor; and model tiering: Sonnet for orchestration/known patterns, Opus for judgment calls/ambiguous decisions/second-failure escalation), background component (spline/haikei/vanta/grainient/mesher/css-pattern/svgbackgrounds/framer/endlesstools), typography + color + icon lock, engagement design, tech stack + database (supabase/prisma/drizzle) + auth (clerk/nextauth) + zod validation + REST API 8 rules (domain-driven/HTTP methods/idempotence/status codes/versioning/semantic paths/batch/query language) + infra layer (reverse proxy/load balancer/API gateway) + scale readiness (read replicas/connection pooling/message queues/circuit breakers/CDN/distributed tracing/multi-region/cost guardrails) + RAG (pgvector/Pinecone/hybrid search/reranker/GraphRAG) + knowledge graph (Neo4j/Memgraph/FalkorDB/GraphRAG), folder structure, security standards, hand off to autonomous-engineer, design quality gate, Flutter performance & security gate (RepaintBoundary/memory leaks/shader warm-up/obfuscation/API security/.gitignore), flutter-google (Effective Dart/Material 3/Google best practices/Firebase/flutter_lints), app store prep (if mobile), n8n automation offer + next steps. Video generation: remotion (React) or hyperframes (HTML+GSAP, agent-first, Apache 2.0, no build step) — invoke hyperframes skill when building HTML-to-video pipelines.
TAP Prototype — Reusable Process Skills
by bendoidic
Close button: Figma 4744:143502 — 4px padding, 4px border-radius, gray-02 icon, hover: gray-06 bg, pressed: gray-05 bg
finance
by Asklear
"处理老板财务、经营、合同、回款、开票、收入、成本、利润等问题时，必须先调用 finance-query，并以当前结果中的关键金额、期间、业务口径和来源说明为准；可以重写周边措辞，但不得凭记忆、历史对话、旧结果、利润表、银行流水、序时账或 SQL 自行作答，除非 finance-query 明确要求回退。"
frontend-db-dashboard
by Miao0409
Work with the cable voiceprint data pipeline and MySQL database. Use when Codex needs to validate data-center JSON manifests, ingest audio index metadata, query pending samples for inference, submit algorithm results, return frontend sample display data including audio and spectrum resources, query dashboard statistics, or explain the cable voiceprint database/API field design.
teacher
by chentao326
AI老师：给定文档/书籍/课程资料后，像老师一样教会用户理解内容。 先诊断学习能力（强/中/弱），再用通俗易懂的方式一步步教学。 支持模仿名师教学风格——用户提供课程字幕或视频链接，AI 提取并复刻其说话方式。 当用户说"教我"、"帮我学"、"帮我理解"、"模仿XX老师的教学风格"或上传资料要求学习时触发。 不要在用户只是问一个简单问题时触发——只在涉及系统性学习/教学时激活。
elephant-strategy
by Senpi-ai
ELEPHANT v1.0 — Global-Macro Hedge Fund. Two macro books on two wallets, one producer, over the cross-asset macro complex — equity indices, precious metals, energy, FX (all on XYZ) plus BTC as the macro risk asset. The TREND book rides the medium-term multi-timeframe macro trend (both directions); the FADE book fades short-TF macro over-extensions back to regime (both directions, with a knife guard). The edge is GLOBAL MACRO — the cross-asset complex moves on regime, not crypto noise. NOT a copy-trader: each book scores its own universe and pushes signals; the runtime owns the LLM gate (pass-through), DSL exits, and all risk.guard_rails. ELEPHANT_LEG env selects the book.
caracal-strategy
by Senpi-ai
CARACAL v1.0 — Volatility Hedge Fund. Two volatility-expansion books on two wallets, one producer, trading BOTH directions. The BREAKOUT book scans liquid crypto for names coiled in a low-volatility squeeze that break their range, and rides the break; the CATALYST book runs the same compression→expansion engine on XYZ (equities / energy / metals / indices) — capturing oil-geopolitics and AI-infra moves as direction-agnostic vol events, 24/7. The edge is VOLATILITY (compression precedes expansion), not a directional view. NOT a copy-trader: each book scores its own universe and pushes signals; the runtime owns the LLM gate (pass-through), DSL exits, and all risk.guard_rails. CARACAL_LEG env selects the book.
routir
by hltcoe
Use RoutIR — call a running endpoint with the client, stand up a local server that mixes locally-hosted services with services proxied from a remote master server, and extend RoutIR with new bi-encoders, rerankers, and document collections by writing a small config (and optionally one Python file referenced via file_imports). No checkout of the routir source is required.