Awesome Skills

English | 简体中文

A concise, curated list of useful Agent Skills and companion tools.

Send this to your AI:

Install this skill: https://github.com/ningzimu/awesome-skills/tree/main/skills/awesome-skills

After that, when a new task comes up, it checks this list and decides whether there is a useful skill to install and use. You do not need to browse the list or compare options every time.

• Skill Ecosystem
• Coding & Development
• Research & Intelligence
• Automation & Productivity
• Obsidian
• Writing & Publishing
• Multimodal & Vision
• Image & Visual Design
• Video & Motion
• Presentations

• Awesome Skills — A meta-skill that queries this live list to automatically discover and install specialist capabilities as tasks require them.

• yusufkaraaslan/Skill_Seekers — Converts documentation websites, GitHub repositories, OpenAPI specs, and PDFs into structured SKILL.md packages with conflict detection.

• anthropics/skills — Anthropic's official collection of example skills, architectural templates, and the foundational Agent Skills specification.

• muratcankoylan/Agent-Skills-for-Context-Engineering — A specialized skill suite for context engineering, dynamic prompt budget optimization, and multi-agent production architectures.

• SamurAIGPT/Generative-Media-Skills — A multimodal media generation suite connecting coding agents to MuAPI models for image, video, audio, and social content synthesis.

• vercel-labs/skills — An open command-line utility (npx skills) for discovering, installing, and managing agent skills across projects.

• limin112/min-skill — A hands-on Claude Code skill collection spanning narrated explainer-video production, WeChat article publishing, and creative handwriting-to-3D story animation workflows.

• multica-ai/andrej-karpathy-skills — Reusable coding guidelines inspired by Andrej Karpathy's observations to curb LLM mistakes: Think Before Coding, Simplicity First, Surgical Changes, and Goal-Driven Execution.

• tanweai/pua — A high-agency coding workflow leveraging corporate PUA / PIP pressure tactics and structured debugging checklists to prevent agents from giving up prematurely.

• oil-oil/vibe-hub-skill — A Vibe Coding terminology assistant that rewrites vague requests into precise agent instructions, clarifies concepts in context, and links to the VibeHub knowledge base.

• vostride/agent-qa — Authors natural-language web and mobile regression tests, debugs broken flows, and triages run evidence via Agent QA's CLI and MCP server.

• rorkai/app-store-connect-cli-skills — Automates App Store Connect workflows via the asc CLI, covering Xcode builds, signing, TestFlight, metadata, screenshots, review submissions, and notarization.

• UI Skills — A design-engineering skill catalog for UI craft, accessibility, motion, and performance, accessible to agents via CLI and MCP.

• Vercel Web Design Guidelines — Audits web UI against modern interface guidelines, accessibility standards, and UX best practices.

• UI UX Pro Max — A searchable design-intelligence library covering UI styles, product palettes, font pairings, UX guidelines, and stack-specific component patterns.

• emilkowalski/skills — A design-engineering skill collection focusing on UI polish, micro-interactions, fluid motion, React Native/Expo, Apple design, and modern Swift.

• twostraws/SwiftUI-Agent-Skill — Reviews and refactors modern SwiftUI code across API adoption, data flow, navigation, accessibility, performance, and best practices.

• Dimillian/Skills — An Apple-platform development skill collection for iOS simulator debugging, Swift concurrency, SwiftUI patterns, performance audits, view refactoring, and App Store release notes.

• greensock/gsap-skills — Official GSAP skills that teach AI coding agents animation best practices, performance optimization, and plugins like ScrollTrigger, MotionPath, and Flip.

• pbakaus/impeccable — A design workflow for AI coding agents that guides UI planning, design critique, visual polish, and live in-browser iteration.

• Leonxlnx/taste-skill — An anti-slop frontend design skill that guides agents in visual direction, landing page aesthetics, and turning design references into distinctive code.

• github/spec-kit — GitHub's official toolkit and templates for adopting Spec-Driven Development with AI coding agents.

• Fission-AI/OpenSpec — An open standard and workflow for Spec-Driven Development, keeping AI coding assistants aligned with specifications and change proposals.

• obra/superpowers — An agentic software development methodology and skill framework that enforces systematic planning, test-driven iterations, and verification.

• mattpocock/skills — A modular software-engineering skill suite covering requirements clarification, ticket decomposition, TDD implementation, parallel code review, and session handoffs.

• Panniantong/Agent-Reach — Gives AI agents access to browse and search 17+ platforms—including X, Reddit, Bilibili, and Xiaohongshu—via CLI and MCP without API keys.

• Jane-xiaoer/x-fetcher — Fetches public X (Twitter) posts, long-form X Articles, engagement metrics, and WeChat Official Account articles for agent workflows.

• mvanhorn/last30days-skill — Researches real-time discussions across Reddit, X, YouTube, Hacker News, and Polymarket, synthesizing grounded recent-trend briefs.

• AI HOT — Aggregates breaking AI industry developments and delivers structured daily intelligence briefings without requiring API keys.

• LearnPrompt/ai-news-radar — A forkable 24/7 AI news radar featuring source quality scoring, deduplication, story clustering, a live dashboard, and agent skills for briefings.

• browserwing/browserwing — Records manual browser interactions directly into deterministic MCP commands or Claude Skills for token-efficient browser automation.

• HumanLayer show-me — Empowers agents to explain complex architectures and codeflows visually using diagrams, pseudocode, diffs, and standalone HTML artifacts.

• kepano/obsidian-skills — Official Obsidian Agent Skills enabling agents to read, write, and manage vaults via the Obsidian CLI, Markdown, Bases, and JSON Canvas.

• axtonliu/axton-obsidian-visual-skills — Generates native Obsidian Canvas boards, Excalidraw sketches, and Mermaid diagrams directly from unstructured text.

• blader/humanizer — Detects and eliminates telltale signs of AI generation from English writing to make text feel human-authored.

• op7418/Humanizer-zh — A localized Chinese skill that strips formulaic tropes, superficial rhetoric, and robotic cadences from AI-written text.

• KKKKhazix/human-writing — A Chinese writing and revision skill distinguishing factual from creative narrative, shaping authentic voice, pacing, and structure.

• jackwener/wechat-article-to-markdown — Scrapes WeChat Official Account articles and parses them into clean, well-formatted Markdown for archival and knowledge bases.

• imraywang/wewrite — An end-to-end WeChat publishing pipeline spanning trending topic discovery, drafting, typography formatting, and one-click draft sync.

• JimLiu/baoyu-skills — A versatile visual content creation suite for designing Xiaohongshu carousel cards, cover graphics, infographics, and article illustrations.

• white0dew/XiaohongshuSkills — Automates Xiaohongshu note publishing, comment interaction, and keyword discovery across compatible coding agents.

• leeguooooo/Mailbox — Enables agents to search, read, triage, and send emails across Gmail, Outlook, QQ, and 163 via CLI or MCP.

• op7418/Claude-to-IM-skill — Bridges Claude Code and Codex to Telegram, Discord, and Feishu/Lark, enabling mobile, remote-controlled coding sessions.

• liustack/modlens — Equips text-only coding agents with multimodal perception, translating pasted images, files, and URLs into structured OCR, layout, and semantic JSON.

• QwenLM/Qwen-MM-Plugins — Official Qwen multimodal skills and MCP servers spanning image/video/document/3D understanding, spatial grounding, media generation, and CAD workflows.

• img2threejs/img2threejs — Reconstructs reference images into clean, procedural, and animation-ready Three.js 3D models with strict quality verification gates.

• tt-a1i/archify — Generates verifiable architecture, sequence, and workflow diagrams as interactive standalone HTML, supporting typed JSON sources and rich media exports.

• cathrynlavery/diagram-design — Designs publication-grade architectural and infographic diagrams across 38 visual formats, outputting clean HTML and SVG without Mermaid templating.

• ningzimu/handdrawn-tech-illustrations — Generates distinctive Chinese hand-drawn technical illustrations, visual concept explainers, and editorial cover art.

• helloianneo/ian-xiaohei-illustrations — Transforms abstract arguments, workplace dilemmas, and architectural concepts into minimalist hand-drawn illustrations starring the Xiaohei IP.

• op7418/guizang-social-card-skill — Generates Swiss-style Xiaohongshu carousels and matched WeChat 21:9 and 1:1 cover sets using modular editorial layouts.

• ningzimu/codex-gpt-image — Integrates GPT Image generation and reference editing into compatible coding agents via Codex authentication without requiring API keys.

• freestylefly/awesome-gpt-image-2 — A comprehensive prompt engineering framework, template catalog, showcase, and reusable skill for GPT Image generation.

• heygen-com/hyperframes — Enables agents to build web-native video compositions using HTML, CSS, and JavaScript, rendering deterministically to MP4.

• hypit-ai/hypit — A video-production framework that transforms creative briefs or reference clips into modular workflows with word-aligned captions and automated compositing.

• calesthio/OpenMontage — An agentic end-to-end video pipeline providing tools and skills for topic research, scriptwriting, asset generation, editing, and final assembly.

• Pluviobyte/rnskill — A Chinese-first interconnected skill suite for coding agents, orchestrating end-to-end video production across ideation, voiceover, motion, subtitles, and QA.

• remotion-dev/skills — Official skills from Remotion for programmatically building, animating, and rendering React-based video compositions.

• Orkas-AI/Orkas-VideoStudio — A local-first CLI and MCP toolkit enabling agents to compose, edit, and render videos from editable plan.json timeline configurations.

• browser-use/video-use — Orchestrates raw footage into polished deliverables, automating rough cuts, filler-word removal, smart captioning, and color grading.

• Agentchengfeng/chengfeng-videocut-skills — A specialized talking-head video cutting workflow for coding agents, automating rough cutting, speech-to-subtitle sync, B-roll insertion, and human-gated export.

• ChatCut-Inc/agent-plugin — Connects coding agents to ChatCut via MCP, enabling multitrack timeline editing, motion graphics, AI asset generation, and in-editor visual checks.

• 0xsline/OpenChatCut — An open-source, local-first AI video editor featuring professional multitrack timelines, MCP control, editable motion graphics, and Remotion export.

• liuluhaixiu/DaVinci-AutoEdit-Agent — Drives DaVinci Resolve via MCP and Python APIs to analyze footage, generate editing blueprints, and construct professional studio timelines.

• JimLiu/baocut — Streamlines speech transcription, multilingual subtitle translation, and clip trimming through BaoCut automation.

• hand-drawn-canvas-animation — Creates editable hand-drawn animated films with JavaScript and Canvas 2D, including varied drawing styles, photo doodles, sand animation, pop-up books, and MP4 rendering.

• adithya-s-k/manim_skill — Instructs agents to programmatically generate precise mathematical, physics, and conceptual explainer animations using Manim.

• Vincentwei1021/video-shotcraft — A production-ready Remotion skill for crafting cinematic product videos with 150+ shot recipes, 2.5D camera moves, beat-synced audio, and JianYing project export.

• geekjourneyx/hyperframes-motion-director — Serves as an AI motion director that adapts articles, READMEs, and web pages into dynamic HyperFrames video stories.

• dexhunter/seedance2-skill — Generates structured, cinematography-grade prompts for ByteDance's Seedance 2.0 video generation model.

• songguoxs/seedance-prompt-skill — Crafts production-ready prompts for Seedance 2.0, covering multimodal reference matching, camera motion replication, and rhythm sync.

• Emily2040/seedance-2.0 — A modular director framework for Seedance 2.0, orchestrating storyboards, multi-shot character continuity, lighting, and failure diagnosis.

• bradautomates/claude-video — Extracts timestamped video keyframes and audio transcripts from local files or URLs, giving agents multimodal ground truth for analysis.

• ningzimu/xiaohu-video-translate — A local pipeline that translates foreign videos into Chinese, automating download, Whisper transcription, contextual subtitle polish, and hardcoded burn-in.

• MapleShaw/yt-dlp-downloader-skill — Automates video and audio extraction across hundreds of platforms using an agent-optimized yt-dlp workflow.

• chuspeeism/dashi-ppt-skill — Generates browser-editable presentation decks across custom themes, with one-click export to HTML, PDF, and editable PPTX.

• hugohe3/ppt-master — Builds native, object-level editable PowerPoint presentations with vectors, charts, animations, speaker notes, and custom template inheritance.

• ningzimu/codex-ppt-skill — Generates visually unified, design-rich image-based PowerPoint decks directly from articles, reports, or outlines.

• ningzimu/image-to-editable-ppt-skill — Reconstructs slide images, scanned decks, and PDF presentations into fully editable, object-level PowerPoint files.

• ningzimu/awesome-ai-ppt — A comprehensive curated directory of AI presentation generators, PowerPoint automation frameworks, and slide engineering workflows.

• op7418/guizang-ppt-skill — Crafts publication-quality single-file HTML presentations and slide graphic cards with modern editorial design aesthetics.

• helloianneo/ian-handdrawn-ppt — Transforms technical courseware and article outlines into distinctive, full-page Chinese hand-drawn presentation slides.