Skills
97,222 packaged capabilities and playbooks for agents. Install any of them with one command.
Creative & media
Edit, review, audit, or draft standalone prose while preserving the writer's meaning and voice. Use for documentation, markdown, emails, blog posts, PRDs, and other dedicated writing when the user wants clearer, more direct, more natural, or less AI-patterned language, or asks whether a draft contains AI-slop patterns. Do not use for routine conversational replies or code unless the user asks to edit its prose.
Provider-independent motion graphics direction for AI agents producing title sequences, lower thirds, explainer graphics, product UI overlays, social ads, kinetic typography, diagrams, data callouts, brand films, and video post. Use when planning, prompting, storyboarding, specifying, implementing, reviewing, or QAing motion graphics across live footage, generated video, UI capture, data visualization, brand systems, typography, accessibility, claim integrity, runtime handoff, and iteration.
Use when the task involves Office document or standalone image handling such as creating, modifying, or converting PPTX, DOCX, XLSX, Report, or IMG files, and the agent should first check whether officecli supports that workflow before choosing the execution path.
Design Roblox worlds, levels, hubs, zones, biomes, obbies, arenas, and exploration spaces. Use for spatial layout, landmarks, player flow, spawn design, objective placement, onboarding paths, traversal, sightlines, social spaces, performance-aware world layout, and guiding players without text.
Use when an agent works in an open Blender scene through the Scenario for Blender extension's local MCP server (scenario-blender on 127.0.0.1:9876): connecting Claude Code, Cursor, Claude Desktop or Codex to Blender; using the headless scenario_blender command; quoting and generating Scenario images; creating PBR materials, 3D meshes, video, audio or 360 environments from the scene; capturing the viewport or selected meshes as references; placing saved results; using Render Video camera paths; or running execute_python.
AI shader generator for WebGL video effects, transitions, masks, and color grading (LUT / color grading / cinematic / film look). Use when the user wants a video effect (filter / effect), a transition (crossfade / wipe / cube / 3d), a mask (reveal), a zoom / push-in, or a color grade — try the built-in effects (zoom, builtin LUTs) before generating a new shader.
Expert patterns for AI video generation, including text-to-video, image-to-video, video editing, and API integration with Runway, Kling, Luma, Wan, and Replicate. Use when “text to video, video generation, image to video, runway api, kling video, luma dream machine, wan video, animate image, ai video, video-generation, text-to-video, image-to-video, runway, kling, luma, wan, replicate, ai-video” are mentioned.
Use when the user provides a short video platform URL or local video/audio file and wants subtitles/AI summary, or when the user asks to list their own AI Douyin historical tasks. Triggers on v.douyin.com, xhslink.com, bilibili.com, b23.tv, YouTube URLs, local .mp4/.mp3/.wav files, or task history requests
Text-to-Speech (TTS), voice cloning, voiceover, narration placement/sync, and custom sound effects (SFX) generator. Use when the user wants generated speech from text, wants to clone a consented voice from uploaded reference audio, wants to correct a misspoken word or short phrase in recorded speech, wants to add/replace/align narration or voiceover for an existing video/timeline, wants to keep existing voiceover synced after visual retiming edits, needs voice audition/selection, or explicitly wants a newly generated/custom sound effect that is not available in the Sound Effects library.
Use this feature when users want to generate videos, shoot short films, or view available video styles. Triggering keywords: short film, make video, shoot short film, short video, AI video, generate video from story, short drama, narrated video, cinematic video, available video styles.
Generate and edit images using OpenAI's GPT Image v2 via EachLabs. Supports text-to-image (gpt-image-v2-text-to-image) and instruction-based editing (gpt-image-v2-edit). Use when the user specifically asks for GPT Image 2 / OpenAI image generation, or needs high-fidelity photorealism, precise text rendering, or reference-faithful edits.
Use when the user wants a REAL photo — 'find a photo of X', 'search for pictures of Y', 'stock image of Z', 'I need a photo for this slide/article'. Searches Pexels free stock library and downloads images. DO NOT use for AI-generated art ('generate an image of', 'create a picture of', 'make me an illustration') — those go to the image_generate tool or fal-enhanced skill. Key distinction: 'photo' = real stock photography = Pexels; 'generate/create/make an image' = AI art = image_generate/fal-enhanced.
Use when adding post-processing effects like bloom, SSAO, depth of field, or anti-aliasing to a Three.js scene. Prevents the common mistake of mixing Three.js EffectComposer with pmndrs/postprocessing, wrong pass order, or missing RenderPass. Covers EffectComposer, 27+ passes, pmndrs/postprocessing, custom ShaderPass. Keywords: post-processing, EffectComposer, bloom, UnrealBloomPass, SSAO, outline, SMAA, FXAA, postprocessing, effects, render pass, glow effect, blur, cinematic, visual effects.
Procedural geometry compilation for Three.js WebGPU/TSL. Use when a task needs an indexed mesh writer, contour-derived terrain or terraces, profile or branch sweeps, or a BatchedMesh, InstancedMesh, or dynamic-update decision.
Use when writing custom GLSL shaders, creating ShaderMaterial or RawShaderMaterial, defining uniforms, or patching built-in materials with onBeforeCompile. Prevents the common mistake of wrong uniform format, missing needsUpdate on uniforms, or not including required defines. Covers ShaderMaterial, RawShaderMaterial, GLSL, uniforms, ShaderChunk, onBeforeCompile, defines, GLSL3. Keywords: ShaderMaterial, RawShaderMaterial, GLSL, vertex shader, fragment shader, uniform, varying, ShaderChunk, onBeforeCompile, custom shader, write shader, custom visual effect, GPU programming.
Create high-performing thumbnails and cover images optimized for click-through rate. Use when asked to create a thumbnail, design a cover image, generate a YouTube thumbnail, or when the user mentions "thumbnail", "cover image", "CTR", or needs a visual asset for video or course content.
Write and optimize prompts for AI-generated outcomes across text and image models. Use when crafting prompts for LLMs (Claude, GPT, Gemini), image generators (Midjourney, DALL-E, Stable Diffusion, Imagen, Flux), or video generators (Veo, Runway). Covers prompt structure, style keywords, negative prompts, chain-of-thought, few-shot examples, iterative refinement, and domain-specific patterns for marketing, code, and creative writing.
Provider-independent workflow for AI agents producing documentary-style montages, archive-driven timelines, interview-supported sequences, nonprofit or advocacy shorts, historical explainers, and hybrid generated/archive edits. Use for editorial thesis development, research/source logs, fact-checking, archive rights, synthetic reenactment disclosure, chronology, interview selects, lower thirds, generated b-roll direction, music restraint, captions, sensitive subjects, review escalation, delivery variants, and documentary QA.
Turn voice-over scripts into screen-recording scripts (CLI + browser), automatically generate ffmpeg recording scripts, and output MP4. Use when the user says “screen record,” “record a tutorial,” “record a video,” “record a clip,” “screen record,” “record additional footage,” “voice-over script to screen recording,” “browser screen recording,” or “record a webpage,” or mentions recording Claude Code or browser interactions
Cycle through multiple PolyHaven materials on a product and render each variant. Great for showing color/finish options. Trigger when asked to show material variants, swap textures, create color options, show finish comparisons, or render product in different materials.
Generates 9 consistent multi-angle fashion-editorial portraits from a single reference image, with locked identity (face/skin/eyes), preserved hairstyle structure, faithful outfit/accessories, and a unified photography style across all frames. Use when the user says "multi-angle", "multi-angle shots", "九连拍", "多角度九连拍", "9-angle portraits", "fashion lookbook", "model consistency series", "consistent portraits from one photo", "generate 9 angles of this model", or "e-commerce model multi-angle pack". Part of the product-shots ecosystem for cross-border e-commerce apparel and accessory listings.
This skill should be used when working with Strudel.cc, a live-coding music environment. Use when creating musical patterns, drum sequences, melodies, basslines, or generative compositions. The user will always want to run Strudel code in the browser, either by copy-pasting or by providing a clickable URL with the code encoded in base64.
Generate images with Venice. Covers POST /image/generate (Venice-native), POST /images/generations (OpenAI-compatible), GET /image/styles (style presets), request fields (prompt, width/height, aspect_ratio, resolution, quality, cfg_scale, steps, seed, variants, style_preset, style_references, enhance_prompt, safe_mode, hide_watermark, format, return_binary), per-model constraints from GET /models, response formats and headers.
Build a personalized product-demo or sales-walkthrough video in HyperFrames for a specific prospect or account — narrated screen-by-screen, branded to them, ending in a clear next step. Turns a generic demo into a "this was made for me" asset for outbound, follow-up, or deal acceleration. Use when you want to send a tailored video instead of booking another live call.
Foundational skill for AI-powered image generation using Gemini 3.1 Flash Image Preview. Handles text-to-image, image editing, multi-turn refinement, and batch generation. Use directly for one-off images; for posters, ads, thumbnails, social posts, etc. prefer the matching graphic design skill, which layers format-specific rules on top of this pipeline. Trigger phrases: "generate an image", "create an image", "AI image", "Gemini image", "generate with Gemini", "image generation", "product image to graphic".
Turn spoken scripts into Chinese slide images in one click with NotebookLM, and convert PDFs to images for use in Jianying. Use when the user says “生成PPT,” “生成幻灯片,” “做PPT图片,” “幻灯片,” or “PPT,” or mentions background images or visual assets for a spoken script.
Used to create standardized ID photos, change ID photo background colors, or produce single-person portrait photo sets based on the user's original image. Triggered when the user asks for an ID photo, one-inch photo, two-inch photo, passport/visa photo, white/blue/red background ID version, portrait photo, three-panel portrait, or themed/scene portrait. Not applicable to group photos, cartoon-style redrawing, or ordinary light portrait retouching. During execution it reads Meitu credentials, calls the local `meitu` CLI, sends the portrait photo, generation prompt and task parameters to the Meitu OpenAPI, and writes the results to the local output directory.
Designs high-performing ad creatives across Instagram, Facebook, TikTok, LinkedIn, Google (Display / Demand Gen), YouTube, Pinterest, and X/Twitter — locking platform dimensions, safe zones, text overlay policy, character limits, industry visual DNA, ad objective hierarchy, composition patterns, CTA strategy, and prompt sanitation. Use when the user says "create an ad", "design an ad creative", "make an ad", "广告创意", "帮我做个广告图", "Instagram ad", "TikTok ad", "Facebook ad", "Google Display ad", "LinkedIn ad", "YouTube ad", "Pinterest ad", "promotional image", "campaign creative", "ad banner", "投放图", or any platform-specific ad request. Routed from `product-shots` when `asset_type == "ad"` or `is_promotion == True`.
Unified image-generation engine for the product-shots ecosystem. Dispatches to the right model family (OpenAI gpt-image-2 / Gemini gemini-3-pro-image-preview Nano Banana Pro) with one parameterised script for text-to-image and image-to-image. Primary backend: OmniMaaS / Cloubic gateway (https://api.omnimaas.com/v1); also supports any OpenAI-SDK-compatible gateway via fallback env vars. Auto-resizes oversized reference images to prevent edge-proxy timeouts. Use when the user says "generate an image", "make a picture", "create a visual", "render this", "edit this image", "做一张图", "生成图片", "改这张图" — or when invoked as the image-generation backend by another product-shots skill (main-image / detail-page / ad-creative / social-post / multi-angle). NOT a creative-direction skill: takes a prompt + optional reference images, returns a file path.
Designs social-platform-native visuals (Feed, Story, Reel, Carousel formats) for Instagram, TikTok, Facebook, Pinterest, RedNote, LinkedIn, X/Twitter and other social platforms — with platform-correct dimensions, safe zones, industry visual DNA, engagement hooks, and cross-slide consistency. Use when the user says "social post", "Instagram post", "IG post", "TikTok post", "Facebook post", "Pinterest pin", "social carousel", "feed post", "story post", "make a post", "design a social visual", "社媒帖", "做一个社媒图", or any underspecified social-media visual request. Routed from `product-shots` for the organic-social branch (non-ad).
Use when setting up lighting in a Three.js scene: ambient, directional, point, spot, area, or environment lighting. Prevents the common mistake of too many shadow-casting lights, wrong intensity units, or missing environment maps for PBR. Covers all 7 light types, PMREMGenerator, HDR environment maps, physically correct intensity, helpers. Keywords: lighting, AmbientLight, DirectionalLight, PointLight, SpotLight, RectAreaLight, HemisphereLight, environment map, HDR, IBL, scene too dark, add light, realistic lighting. PMREMGenerator, scene.environment.
Representation-first Three.js WebGPU/TSL particles, ribbon or history trails, and effects. Use for analytic or recurrent particle motion, stable-slot or scan-compacted GPU pools, flow-conforming shells and wakes, dissolving debris, or effect-specific HDR and depth integration.
Procedural materials for coherent NodeMaterial PBR systems in Three.js WebGPU/TSL. Use when a task needs shared identities across PBR channels, must select UV, array, atlas, or projected mapping, needs footprint-filtered normals and specular AA, or needs dynamic per-instance material state with explicit output ownership.
Package one video/script/transcript into a complete publish kit: cover titles, platform titles, per-platform tags, description, and creative cover art direction with pre-review before image generation. Covers Xiaohongshu, Douyin, WeChat Channels (视频号), Bilibili, and YouTube. Use when the user asks for video cover design, 封面包装, 视频包装, 发布文案, 标题公式, 封面标题, platform tags for a video, or says '给这期视频做封面/标题/描述/标签'. Keywords: video cover, thumbnail title, publish kit, self-media packaging, 小红书封面, 抖音封面, B站封面.
Plan, write, and publish for YouTube and YouTube Shorts. Use when the user wants high-CTR video titles, an SEO description with chapters and links, a first-30-seconds retention script (or a 3-second Shorts hook), a designer-ready thumbnail brief, a community-tab post or poll, or a weekly upload plan. The user supplies the video; the skills produce the title, description, hook, and thumbnail brief, then on approval upload and schedule via the Publora API.
Load this skill whenever the project contains audio or video content, media players, podcasts, video embeds, or any <audio>/<video> elements. Under no circumstances publish audio or video without captions, transcripts, and audio descriptions where required. Absolutely always apply WCAG 1.2 criteria for time-based media.
Structure the body and close of a short-form video into a reusable Narrative Brief with explicit arc choice, beat sequence, product integration timing, emotional arc, and CTA design. Use after the hook mechanism and viewer question are locked and before storyboard or prompt writing starts.
Use when exporting a model, character or animation from Blender to a game engine, the web or a 3D printer: FBX, glTF/GLB, USD/USDZ or STL for Unity, Unreal Engine, Godot, Roblox, three.js or Apple AR Quick Look. Also when an export arrives at the wrong scale (100x, 0.01), rotated 90 degrees, facing backwards, with extra leaf bones, without its animation, blend shapes or textures; when an FBX loses its metallic or roughness map; when a generated model needs its real size and its pivot on the floor; when making LODs or collision meshes (UCX_, -colonly) for an engine; when converting a folder of models in batch; or when checking an exported file by re-importing it.
Professional Seedance 2.0 platform AI video script and storyboard generator. Use when the user asks to: (1) convert an article/story into a video script, (2) generate Seedance 2.0 storyboard prompts, (3) plan a multi-episode AI video series, (4) create character/scene/prop generation prompts for image models such as GPT-Image-2, Seedream, Nano Banana Pro. Input can be a full novel, an article, or a short story outline. Output includes a complete screenplay in standard script format (△ shot description + dialogue + OS/VO + flashback + subtitles), episode breakdown, asset generation prompts, and a storyboard script in Seedance 2.0 format. Supports the "text-asset-video-edit" workflow, covering the full process of scriptwriting, asset building, video generation, storyboard optimization, and sound design.
Worked qupai-system (曲牌体) opera scripts, act by act. Guan Hanqing's three surviving zaju (窦娥冤／救风尘／金线池 — twelve acts, 126 tunes, one mode and one rhyme per act, sung/spoken ratios); Tang Xianzu's 牡丹亭 (55 scenes, ~420 tunes, 集唐 closing quatrains, the 圆驾 south-north composite suite); Kong Shangren's 桃花扇 (44 scenes plus four unnumbered, the 副末 frame line, upper/lower halves, a fan's seven transformations, his sixteen 凡例); 长生殿 (Wang Jilie's fifty-scene staging table); Wei Minglun's 巴山秀才 (1983 Sichuan 高腔, 43 tunes, nine 帮腔 passages). Every case states its edition and collation status, then structure, tunes, stage conventions and transferable devices. Method is in sw-chinese-opera-qupai; this skill holds the worked scripts. Use when you want a real example of how a 传奇 or 杂剧 is assembled, need actual scene counts and tune counts to imitate, are judging where a famous script breaks prosody, or need models for 副末开场, 集唐下场诗, 南北合套 or 帮腔 (曲牌体全本／牡丹亭／桃花扇／窦娥冤／救风尘／长生殿／巴山秀才／唱段表／结构表).
Use when composing a complete Three.js scene from scratch or when asked to create a 3D visualization. Provides decision trees for choosing lighting setup, materials, camera type, controls, and post-processing pipeline. Orchestrates other Three.js skills. Keywords: create scene, build 3D, Three.js project, scene setup, visualization, product viewer, 3D application, scene composition, how to start Three.js, new 3D project, basic scene.
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.