Skills
97,222 packaged capabilities and playbooks for agents. Install any of them with one command.
Creative & media
Route academic-figure work into the lightest workable PaperBanana mode instead of jumping straight to a full multi-agent generation run. Use when the user needs publication-quality methodology diagrams, statistical plots, figure evaluation, polishing an existing figure, batch/sweep generation, or a full-paper figure package from text or PDF. Even if the user does not say "paperbanana" — also triggers on: academic figure, methodology diagram, publication figure, generate diagram from paper, statistical plot from CSV, figure evaluation, polish figure, NeurIPS/ICML figure, arxiv illustration, plan-then-refine diagram pipeline.
Provide concise, goal-aligned suggestions before generating or optimizing image, video, audio, GPT Image, or storyboard prompts. Use when the user asks for advice, direction, risks, alternatives, or "how should this prompt be improved" before producing the final prompt. This skill gives suggestions only and should not replace the main prompt-building skill.
Produce text-to-speech with Alibaba/Qwen Qwen3-TTS through DashScope/Model Studio or open-weight Qwen3-TTS checkpoints. Use when an agent must choose Qwen3-TTS models, voices, realtime versus non-realtime synthesis, voice cloning, voice design, instruction/prosody controls, multilingual or dialect speech, audio artifact handling, regional routing, pricing/privacy/licensing constraints, or production QA for narration, characters, assistants, audiobooks, ads, localization, or voice-enabled media.
Apply a new brand across existing videos after a rebrand - audit every scene for old colors, logos, fonts, and dated intros/outros, swap them for the new brand kit, and leave narration and timing untouched. Use when the user says "we rebranded, update our videos", "swap the old logo in this video", "apply our new brand colors to this video", "our video library still has the old branding", or "replace the intro with our new one".
Restructure a video's narrative by re-sequencing its sections into a better order and smoothing the narration seams so the new flow sounds intentional. Use when the user says "reorder the scenes", "move the pricing part to the end", "lead with the outcome", "restructure this video", or wants the same material told in a different order.
Turn a Roblox game concept or the current conversation into a short, buildable PRD (game design brief) and publish it to the project tracker. Use when the user wants to lock a chosen concept into a spec before building, write a design brief, or create a PRD for a Roblox game or feature. Pairs with roblox-to-issues, which slices the PRD into vertical slices so the game is built one playable piece at a time instead of all at once.
Use when recreating a trending reel, TikTok or Short with a different performer: the same dance, moves, timing and camera, performed by the user's own character from a reference sheet, optionally in another outfit or a redecorated room, keeping the original song in sync. Keywords: reel recast, trend remake, TikTok dance, replace the dancer, character swap, choreography transfer, video to video, keep the music, Seedance edit mode.
Playwriting for banqiang-system Chinese opera (板腔体：京剧、豫剧、越剧、秦腔、评剧、沪剧、淮剧、川剧弹戏), where sung passages are paired seven- or ten-character lines varied by metre (板式) rather than filled into fixed tune patterns. Covers genre (唱工戏／做工戏／武戏), the main line, technical structure and 排场, how 行当 and a named star decide the writing, 程式, lyric craft (唱词：上下句、重点句、垛句、辙口), 宾白 and the sung/spoken seam, duets, what the writer marks on the page (Peking-opera vs Yue-opera style), the interface with actors and musicians, restoring old plays, and three contemporary routes including 现代戏. Sources; 范钧宏、翁偶虹、刘吉典、樊尚林、张庚郭汉城、罗怀臻、陆军, plus full scripts (锁麟囊／沙家浜／白蛇传／朝阳沟／潘金莲). Use when writing or revising 唱词, choosing 板式, writing for a performer or role-type, adapting a traditional play, writing a 现代戏, judging a project's genre, marking 锣鼓, or drawing the boundary before general dramaturgy skills apply. Not for 元杂剧/传奇/昆曲, which are the qupai system; see sw-chinese-opera-qupai.
Implement production bloom in advanced Three.js scenes. Use for HDR signal ordering, bloom-node controls, dual selective bloom with guaranteed material restoration, scene-relative emissive hierarchy, and effect-isolation diagnostics.
Write one model-agnostic video prompt spec, then compile it to whichever video model you can actually call. Use for cross-model prompt work, model comparison matrices, reusing one brief across providers, or when the target model is not yet available and the work must proceed on another one
Produce, edit, extend, and govern short videos with xAI's direct Grok Imagine Video API. Use for text-to-video, image-to-video, multi-reference video, natural-language video edits, video continuation, exact media-cost approval, asynchronous request recovery, Files/Batch integration, moderation review, and API privacy or residency decisions. Covers `grok-imagine-video` and image-only `grok-imagine-video-1.5`; excludes Grok consumer subscriptions, Grok on X, still-image generation, voice APIs, and third-party gateways.
Use ACE-Step API to generate music, edit songs, and remix music. Supports text-to-music, lyrics generation, audio continuation, and audio repainting. Use this skill when users mention generating music, creating songs, music production, remix, or audio continuation.
Plan, prompt, call, edit, iterate, and productionize Alibaba Cloud Model Studio image generation with current Wan 2.7 Image, Qwen-Image 2.0, and Z-Image Turbo models. Use for Alibaba/DashScope text-to-image, multi-reference generation, instruction editing, character-consistent image sets, typography and layout work, regional authentication, model selection, API integration, cost/rate-limit planning, troubleshooting, safety, rights, privacy, and output QA.
Inspect and validate Blender assets technically and visually. Use for `.blend`, `.glb`, `.gltf`, `.fbx`, or `.obj` quality checks; topology and export review; evaluated triangle/material/hierarchy metrics; standardized multiview renders; fresh-import verification; or evidence-backed review of an agent-generated mesh.
Design a matched pair of branded stingers - a 3-5 second intro (logo moment plus title slot) and a visually matched outro (CTA card) - so every video a team ships opens and closes the same way. Use when the user says "make us a branded intro", "intro and outro for our videos", "logo sting", "opening and closing bumpers", "end card with our CTA", or wants consistent on-brand openers and closers for their video library.
Build and operate production image generation and natural-language image editing with ByteDance Seedream through first-party Volcengine Ark (China) or BytePlus ModelArk (global), including model and region selection, multi-reference and grouped outputs, streaming, secure artifact handling, retries, cost controls, prompting, safety, and rights review.
Split one long onboarding or training recording into a clearly labeled, ordered sequence of chapters cut at clean topic boundaries, each opening on a silent animated title card naming the chapter and its place in the sequence. Delivered as one project with labeled internal segments, or as several standalone per-chapter projects if each module should be individually shareable. Use when the user says "split this onboarding video into chapters", "turn this recording into a structured course", "break this long video into modules", "chapter this training video", or "make a module-by-module version of this demo".
Main entry point for multi-platform content creation. Coordinates the entire creation workflow: research → images → writing → polishing → platform conversion. Use to create articles, write content, do topic research, and publish to Zhihu/Xiaohongshu/WeChat Official Accounts.
Production guidance for converting sourced data and approved claims into truthful, accessible, deterministic animated visualizations with D3. Use for data-driven charts, maps, networks, hierarchies, transitions, annotations, responsive video variants, frame-by-frame browser rendering, and visualization QA; not for generic dashboards, unsourced decorative charts, or ordinary motion graphics without a data contract.
Provider-independent production guidance for AI agents creating or retouching appetizing food and beverage images, recipe videos, restaurant/menu visuals, CPG ads, ecommerce product images, beverage pours, packaging/lifestyle shots, and social food clips, including truthfulness, labeling/claims discipline, styling, prompt translation, platform variants, accessibility, rights, alcohol, and QA.
Use Google Lyria music generation models through Google Cloud / Vertex AI / Gemini Enterprise Agent Platform for production music beds, songs, vocal tracks, image-conditioned music, short-form social audio, ad music, and video scoring. Covers Lyria 2, Lyria 3 Clip, and Lyria 3 Pro model selection, prompt construction, rights/privacy checks, SynthID/C2PA provenance, artifact custody, and QA.
Produce Hedra character, talking-avatar, lip-sync, motion-avatar, and live avatar work. Use when an AI agent needs to choose Hedra models, prepare image/audio/script inputs, direct character performance, call or plan around the Hedra API or Studio, evaluate avatar/lip-sync results, handle consent/rights/privacy, or troubleshoot Hedra character video production.
Still-to-video conversion guide: model selection, motion prompting, and camera movement. Covers Wan 2.5 i2v, Seedance, Fabric, Grok Video with when to use each. Use for: animating images, creating video from stills, adding motion, product animations. Triggers: image to video, i2v, animate image, still to video, add motion to image, image animation, photo to video, animate still, wan i2v, image2video, bring image to life, animate photo, motion from image
Split a recorded live training session or webinar into short, self-contained, on-demand modules a viewer can watch individually later. Unlike a scripted recording, a live session needs real cleanup first: dead air, "can everyone see my screen" setup chatter, off-topic tangents, and sprawling Q&A get identified and cut before any module boundaries are drawn, not just chopped at topic marks. Use when the user says "turn this recorded training session into on-demand clips", "split this live webinar into modules", "make this recorded session watchable in pieces", "cut this live training into short videos", or "clean up this webinar recording into standalone modules".
Use MiniMax's first-party Hailuo video API safely and reproducibly across global and mainland-China platforms. Covers current text-to-video, image-to-video, first/last-frame, and subject-reference models; prompt and camera craft; region/account routing; exact cost approval; asynchronous jobs and callbacks; artifact provenance; and rights, likeness, disclosure, and data-governance review. Use when designing, implementing, debugging, or evaluating direct MiniMax/Hailuo video-generation API workflows. Do not use for the consumer Hailuo site, Video Agent templates, gateways, or unrelated MiniMax modalities.
AI animation sprite generation. From a one-line text description, create a character + motion animations (walking, running, attacking, magic, and about 100 other actions) + an 8-direction sprite set, and export game-engine-ready bundles (sprite sheet · manifest.json · Aseprite JSON · per-state GIF/APNG · individual frame PNGs). Use when the user wants to generate game sprites, character animations, sprite sheets, sprite atlases, or 8-direction sprite sets from a text description. Triggers on: perfectpixel, ppgen, sprite generation, character animation, sprite sheet, sprite atlas, 8-direction sprite, god-tibo-imagen sprite, gemini sprite.
Split one onboarding recording into a multi-part video sequence sized and paced for embedding in separate onboarding drip emails: email 1 covers setup, email 2 the core workflow, email 3 an advanced tip. Each part is promoted to its own standalone project with a short animated framing card so it works whether or not the recipient watched an earlier email. Clueso does not send email or manage a drip sequence; it produces the video clips only, and the user wires them into their own email service provider. Use when the user says "split this recording into an onboarding email video series", "turn this into a multi-part email video sequence", "make short clips for our onboarding drip campaign", or "cut this demo into parts for our email drip".
From one screen recording, always produces both a polished video and a full step-by-step help article: two equal, always-built deliverables rather than one primary format with the other optional, with the article built as a real, live article inside a Clueso project. Use this instead of recording-to-help-article specifically when the team needs both formats by default. If the user already has an existing Clueso article, the finished video and written steps can be attached to and merged into that one instead of a new project. Use when the user says "turn this recording into a video and a help doc", "I need both a video and written steps from this capture, no matter what", "make a guide and a video from this recording, I want both every time", or "give me a polished video plus documentation from this screen recording".
Turn one finished video into a channel kit - a captioned, resized social cut of the strongest 15–30 seconds, a looping GIF of the key interaction for newsletters and in-app embeds, and the embed-ready full version. Use when the user says "cut this for LinkedIn", "make a social version of this video", "turn this demo into a GIF for the newsletter", "repurpose this video for our channels", or "I need short cuts of this for socials and the changelog".
Clone an existing 16:9 Clueso video into a vertical (9:16) or square (1:1) canvas for Reels, TikTok, Shorts, or feed posts, then reflow every element (backgrounds, text, logos, cursors and spotlights) so nothing crops badly, floats off-frame, or points at the wrong spot after the resize. Use when the user says "repurpose this explainer as a vertical short", "turn this into a Reel", "make a TikTok version of this demo", "resize this for Instagram Stories", "give me a square version for the feed", or "I need this in 9:16".
Use the RPGJS Studio HTTP API to create or manage a 2D RPG game. Trigger this skill when Codex needs to CRUD maps, map events, database records, media assets, or general project settings in RPGJS Studio, especially when the task should be done through `curl` or another HTTP client with an API key and configurable base URL.
Build and operate production-safe Runway API video generation, video editing, and character-performance workflows with native Runway models, explicit paid-call approval, duplicate-create protection, asynchronous task handling, secure media transfer, and evidence-aware governance.
Generate online course and coaching program promotional video prompts for Seedance 2.0 on Higgsfield. Use for course trailers, coaching ads, educational content promos, masterclass teasers, webinar promotions, or any educational product video. Triggers on course, coaching, masterclass, webinar, tutorial, education, teaching, training, class, program, academy.
Generate customer testimonial and social proof video prompts for Seedance 2.0 on Higgsfield. Use for customer stories, case study videos, review showcases, social proof content, success story videos, or any content featuring customer results and experiences. Triggers on testimonial, review, case study, social proof, customer story, success story, results, feedback, client.
Turn an uploaded slide deck (PPT) into a narrated video that doesn't feel like a deck: slide text rewritten into spoken narration instead of read aloud, slides that don't earn screen time cut or merged, motion added to the slides that matter, and clean transitions at section boundaries. Use when the user says "turn this deck into a video", "convert my PowerPoint to video", "narrate these slides", "make a video from this presentation", or uploads a deck and wants a video version.
Worked full-length scripts of banqiang-system Chinese opera with their creation records (板腔体全本语料库). Primary texts; 翁偶虹《锁麟囊》 (the revised score edition, 65 arias, 15 self-account-vs-score checks), 汪曾祺等《沙家浜》 (the 1965 text, not the 1970 one; 36 arias, a 12-row two-version diff), 田汉《白蛇传》 (the 1955 book, not the web revision; 16 scenes, 王瑶卿说腔五条), 杨兰春《朝阳沟》 (1978 revival, 50 arias, 口语入唱, per-scene sung/spoken ratios), 魏明伦《潘金莲》 (跨时空, missing pages), 罗怀臻's 越剧《李清照》《真假驸马》, 淮剧《金龙与蜉蝣》, 甬剧《典妻》; plus index tables of 翁偶虹's 60 创作记 and 范钧宏's seven 改编札记. Use when you want to see how a 唱工戏 lays out its 板式, need real scene counts, aria counts and 辙口 to imitate, want the road from raw material to 审腔, are writing a 现代戏 or colloquial lyrics, or compare one author across two声腔 systems. Method is in sw-chinese-opera-banqiang; this skill holds the worked scripts and their creation records.
Chinese opera playwriting for the qupai (曲牌体) system — 元杂剧, 明清传奇, 昆曲, 川剧高腔, where sung text is filled into fixed tune patterns (填词) rather than built from paired lines varied by metre. Covers stage vs study (李渔 vs 吴梅/王季烈), the 清曲/剧曲 switch, the zaju branch (四折一楔子, one 宫调 and one rhyme per act, 一人主唱, 题目正名), the chuanqi branch (出, 副末开场, 南北合套, 集唐下场诗), tune-filling (格律, 前腔/换头/幺, 衬字, 务头, 尾声, 四十禁), the seventeen 宫调 and their affects, 套数, 排场 alternation, the spoken-to-sung cue, 帮腔, adaptation, and modern practice. From 李渔《闲情偶寄》, 吴梅《顾曲麈谈》《曲学通论》, 王季烈《螾庐曲谈》, 孔尚任's 凡例, 关汉卿, 汤显祖, 魏明伦. Use when writing or revising a 曲牌体 script, choosing a 宫调 for an emotion, laying out a 套数, filling one tune pattern, writing a zaju act or a chuanqi scene, cutting an old 传奇 into a playable version, writing a modern qupai opera, or judging whether a famous script's prosody may be copied. Not for 京剧/豫剧/越剧, which are the banqiang system; see sw-chinese-opera-banqiang.
Implement the Syncfusion ASP.NET Core Image Editor (EJ2) component. Use this when adding image editing to an ASP.NET Core application, including cropping, rotating, annotations, fine-tuning (brightness, contrast, blur), and filter application. Covers opening/saving images, toolbar customization, redaction, resizing, and image editor event handling with ejs-imageeditor.
Use meitu-cli to call a standalone LLM to generate text/copy/storyboards/video prompts from scratch. Trigger when the user mentions writing copy, polishing, rewriting, slogans, product descriptions, social media copy, articles, storyboards, storyboard scripts, or prompt optimization. Disabled when the context already contains complete content; use write_file instead.
Build well-crafted production procedural meshes in Three.js. Use for complete hard-surface assemblies and humanoid robots, profile extrusion, parameter-curve and spine lofts, pillow panels, exact polygon cuts, inset, revolve, sweep, solidify, bevels and fillets, shell thickness, direct-topology apertures, semantic mesh writers, or diagnosing primitive-built forms, coplanar flicker, loose/non-manifold geometry, detached parts, interpenetration, support, clearance, and swept-envelope defects.
Generate authored procedural trees, grass, ivy, flowers, and vegetation in Three.js or raw WebGPU. Use for surface-following vines, painted ivy paths, stylized or GPU grass, GPU-culled virtual flower fields, trunks, recursive branches, roots, canopies, leaf cards, species presets, deterministic growth, distance-tiered plant geometry, and rooted blade, stem, or petiole-hinge wind.
Bring a product UI to life from a screenshot or mockup: an animated scene of the interface with a choreographed interaction - cursor journey, panels sliding in, states changing - plus captions or tight narration naming the benefit. Use when the user says "animate this screenshot", "animate this mockup", "make our UI move", "UI animation for this feature", "turn this design into a demo animation", or shares a product screenshot/mockup and wants a short polished clip showcasing an interaction.
Use meitu-cli to do outward video canvas outpainting, with AI intelligently filling in the extended areas based on the prompt. Trigger when the user mentions extending a video, video outpainting, expanding the frame left/right/up/down, video outpainting, adding imagery to a video, expanding more background, or making the video frame wider.
Generate and edit production images with xAI's first-party Grok Imagine API, including model selection, multiple references, synchronous and Batch API workflows, durable output handling, prompting, iteration, QA, cost and rate controls, privacy, safety, and rights review. Use for direct xAI image API integrations, not Grok video or third-party gateways.
Use Adobe Firefly Services to generate, edit, expand, fill, match, composite, and upscale still images through the current REST APIs. Apply when selecting Firefly Image 5 versus Image 3/4 or custom models, implementing authenticated asynchronous image workflows, using style or structure references, building product composites, handling provenance and enterprise rights, or diagnosing Firefly image jobs in production. Exclude Firefly video, audio, and unrelated Creative Cloud APIs.
WeChat official account cover | WeChat official account images | WeChat official account illustrations | AI image generation — AI cover and image generation for WeChat official accounts, automatically matching an art style to the article title and content, multiple options per draft, reusable style presets. For WeChat official account editors, self-media, and brand design. Trigger words: 「封面」「配图」「插图」「生成图片」「给文章加图」「做个封面」「文章插图」「配个图」. To publish only a set of images without writing the body text, use aws-wechat-sticker; for a multi-step pipeline (write + review + layout + images + publish), use aws-wechat-article-main.
Fans out one finished Clueso video into parallel language variants: duplicating the source project per target language, swapping in a translated script and matching voice, re-syncing durations, and exporting all of them in one pass without ever touching the original. Use when the user says "export this in English, Spanish, German, French, and Portuguese", "batch dub this video into 5 languages", "give me localized versions of this launch video", "auto-dub this for our EMEA markets", or "I need this same video in multiple languages, fast".
Split a long video into titled chapters with a clean title card at each boundary. Use when the user says "add chapters", "chapterize this", "break this into sections with titles", "add section title cards", or shares a long training or walkthrough video that runs as one unbroken stream.
Produces every required language version of a compliance training video from one source recording, applying a locked glossary and extra scrutiny to legally load-bearing lines (required disclosures, regulatory terms) rather than paraphrasing them loosely. Layers compliance-specific accuracy and traceability discipline on top of the duplicate, voice, narrate, re-sync, export mechanics used for general localization. Does not replace legal or compliance sign-off on the translated content. Use when the user says "localize this compliance training into all our required languages", "translate this mandatory training video for every region we operate in", "produce the French and German versions of this compliance course", "we need this harassment training dubbed for every country we operate in", or "get this policy video ready in our required languages".
Assists in writing high-quality content by conducting research, adding citations, improving hooks, iterating on outlines, and providing real-time feedback on each section. Transforms your writing process from solo effort to collaborative partnership.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.