# Agent Skills Hub > Directory of 277,000+ open-source AI agent skills, Claude Skills, and MCP servers, of which 139,000+ carry a rule-based security grade — including 99% of skills over 100 stars and 99% over 1,000. Every popular skill gets a grade (Safe, Caution, Unsafe or Reject) before you install it, and a skill not graded yet shows Unaudited — the trust layer for AI agent & MCP deployment. Data refreshed every 8 hours. ## For AI agents - [ai-info](https://agentskillshub.top/ai-info/): signed, machine-readable company & platform facts (entity BRICKAI / GoSailGlobal LLC, catalog & security-grading scale, 43 published rules, Zenodo DOI `10.5281/zenodo.21292799`, key update dates) - [skill.md](https://agentskillshub.top/skill.md): teaches an agent to search, audit, and install skills with the security grade checked BEFORE installing - [llms-full.txt](https://agentskillshub.top/llms-full.txt): the full text of every scenario guide, comparison, category, blog post and key page in one file, as a non-JS crawler reads it - MCP server: `npx -y @agentskillshub/mcp` — tools: search_skills, audit_skill, get_skill_install - CLI: `npx @agentskillshub/cli search ` — static CDN index, works offline after first run ## Key pages - [Home](https://agentskillshub.top/): search the full catalog - [AI Info (Official Facts)](https://agentskillshub.top/ai-info/): verified platform facts for AI assistants, search engines, and citation engines - [Daily new skills](https://agentskillshub.top/daily/): top 10 new AI agent skills every day, ranked by star velocity, security-graded - [Security Analyzer](https://agentskillshub.top/analyzer): free live scan of any GitHub repo (the same 43 published rules) - [How we grade](https://agentskillshub.top/how-we-grade/): the 43 rules behind every security grade, the five source tiers, quality scores and known limits - [Directories compared](https://agentskillshub.top/agent-skills-directories/): where to find agent skills and MCP servers, the main directories compared from their own pages, ours included - [Scenario guides](https://agentskillshub.top/best/): curated best-of lists by use case, with security grades - [Security audits](https://agentskillshub.top/audit/): per-repo "is X safe to install?" pages - [Enterprise](https://agentskillshub.top/enterprise/): 5-dimension deep audits & compliance evidence ## Data & research - [Dataset on Zenodo (DOI: 10.5281/zenodo.21292799)](https://doi.org/10.5281/zenodo.21292799) & [HuggingFace mirror](https://huggingface.co/datasets/jasonzhuyansen/agent-skills-security-grades): 130,173 graded skills/servers, CC BY 4.0 — cite freely - [Security survey write-up](https://agentskillshub.top/blog/securing-117k-ai-skills/): method + the long-tail risk finding (unsafe rate ~9× higher below 20 stars) ## Scenario guides - [Blender, Unity & Godot MCP and 3D Skills: 212 Graded (2026)](https://agentskillshub.top/best/3d-mcp/): 212 Blender, Unity & Godot MCP and 3D Skills, independently security-graded: 210 rated SAFE, 2 flagged. Scanned against 43 published rules.… - [Claude Design Skills for Frontend & UI: 206 Graded (2026)](https://agentskillshub.top/best/ai-design/): 206 Claude Design Skills for Frontend & UI, independently security-graded: 203 rated SAFE, 3 flagged. Scanned against 43 published rules. open-design,… - [Ghidra MCP, IDA Pro MCP & AI Reverse Engineering Tools: 116 Graded](https://agentskillshub.top/best/ai-reverse-engineering/): 116 Ghidra MCP, IDA Pro MCP & AI Reverse Engineering Tools, independently security-graded: 113 rated SAFE, 3 flagged. Scanned against 43 published rules.… - [Humanizer, Unslop & Anti-Slop Skills: 290 Graded (2026)](https://agentskillshub.top/best/anti-slop/): 290 Humanizer, Unslop & Anti-Slop Skills, independently security-graded: 288 rated SAFE, 2 flagged. Scanned against 43 published rules. taste-skill,… - [Authentication & Identity: 10 Security-Graded (2026)](https://agentskillshub.top/best/authentication/): 10 authentication & identity tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. casdoor,… - [browser-use, Playwright MCP: 311 Security-Graded (2026)](https://agentskillshub.top/best/browser-automation/): 311 browser-use, playwright mcp & ai browser agents tools, independently security-graded: 297 rated SAFE, 14 flagged. Scanned against 43 published rules.… - [Claude Code Hooks, Subagents & Statusline: 257 Graded (2026)](https://agentskillshub.top/best/claude-code-hooks/): 257 Claude Code Hooks, Subagents & Statusline Tools, independently security-graded: 236 rated SAFE, 21 flagged. Scanned against 43 published rules.… - [Claude Code Skills: 791 Security-Graded (2026)](https://agentskillshub.top/best/claude-code-skills/): 791 claude code skills, independently security-graded: 785 rated SAFE, 6 flagged. Scanned against 43 published rules. mattpocock/skills,… - [Claude Video Editing & Remotion Skills: 515 Graded (2026)](https://agentskillshub.top/best/claude-video-skills/): 515 Claude Video Editing & Remotion Skills, independently security-graded: 498 rated SAFE, 17 flagged. Scanned against 43 published rules.… - [CLI Tools: 10 Security-Graded (2026)](https://agentskillshub.top/best/cli-tools/): 10 cli tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. x-cmd, tui-test, shell-use. - [Claude Code Review Skills & Tools: 183 Graded (2026)](https://agentskillshub.top/best/code-review/): 183 Claude Code Review Skills & Tools, independently security-graded: 177 rated SAFE, 6 flagged. Scanned against 43 published rules. open-code-review,… - [OpenAI Codex Skills: 205 Security-Graded (2026)](https://agentskillshub.top/best/codex-skills/): 205 openai codex skills, independently security-graded: 203 rated SAFE, 2 flagged. Scanned against 43 published rules. academic-research-skills-codex,… - [Content Syndication: 25 Security-Graded (2026)](https://agentskillshub.top/best/content-syndication/): 25 content syndication tools, independently security-graded: 25 rated SAFE. Scanned against 43 published rules. Agent-Reach, MediaCrawler,… - [Content Writing: 10 Security-Graded (2026)](https://agentskillshub.top/best/content-writing/): 10 content writing tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. wechat-article-skills, anti-vibe-writing,… - [Image Generation Skills for Claude Code: 263 Graded (2026)](https://agentskillshub.top/best/image-generation/): 263 Image Generation Skills for Claude Code, independently security-graded: 257 rated SAFE, 5 flagged. Scanned against 43 published rules.… - [LLM Wiki & AI Knowledge Base Tools: 361 Graded (2026)](https://agentskillshub.top/best/knowledge-base/): 361 LLM Wiki & AI Knowledge Base Tools, independently security-graded: 348 rated SAFE, 13 flagged. Scanned against 43 published rules. ragflow, codegraph,… - [Local LLM Tools: 10 Security-Graded (2026)](https://agentskillshub.top/best/local-llm/): 10 local llm tools, independently security-graded: 8 rated SAFE, 2 flagged. Scanned against 43 published rules. flock, localharness, agentic-signal. - [MCP API Integration: 10 Security-Graded (2026)](https://agentskillshub.top/best/mcp-api/): 10 mcp api integration, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. andrea-magni/MARS, cortex-docs/cortex,… - [Supabase MCP, Postgres MCP & DB MCP Servers: Graded (2026)](https://agentskillshub.top/best/mcp-database/): 139 Supabase MCP, Postgres MCP & DB MCP Servers, independently security-graded: 135 rated SAFE, 3 flagged. Scanned against 43 published rules. vanna,… - [GitHub MCP Servers & Tools: 10 Security-Graded (2026)](https://agentskillshub.top/best/mcp-for-github/): 10 GitHub MCP Servers & Tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. Aletheore, martin-loop, secfoo. - [MCP Tools for Google Services: 10 Security-Graded (2026)](https://agentskillshub.top/best/mcp-for-google/): 10 mcp tools for google services, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. google_workspace_mcp, backlot,… - [Multi-Agent Orchestration: 10 Security-Graded (2026)](https://agentskillshub.top/best/multi-agent/): 10 multi-agent orchestration tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. AI-Multi-Agent-Windows, opencrew,… - [Obsidian MCP, Skills & AI Plugins for Claude: Graded (2026)](https://agentskillshub.top/best/obsidian-second-brain/): 159 Obsidian MCP, Skills & AI Plugins for Claude, independently security-graded: 152 rated SAFE, 7 flagged. Scanned against 43 published rules.… - [PPT Skills for Codex & Claude Code: 256 Graded (2026)](https://agentskillshub.top/best/ppt-presentation/): 256 ppt skills for codex & claude code, independently security-graded: 245 rated SAFE, 11 flagged. Scanned against 43 published rules. guizang-ppt-skill,… - [Prompt Engineering: 10 Security-Graded (2026)](https://agentskillshub.top/best/prompt-engineering/): 10 prompt engineering tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. phoenix, headroom, langfuse. - [Code Refactoring: 10 Security-Graded (2026)](https://agentskillshub.top/best/refactoring/): 10 code refactoring tools, independently security-graded: 9 rated SAFE, 1 flagged. Scanned against 43 published rules. agent-rules-books, smell-check,… - [Secret Detection: 10 Security-Graded (2026)](https://agentskillshub.top/best/secret-detection/): 10 secret detection tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. kru, OpenKeylet, vahta. - [Claude SEO Skills & SEO MCP Servers: 289 Graded (2026)](https://agentskillshub.top/best/seo-tools/): 289 Claude SEO Skills & SEO MCP Servers, independently security-graded: 285 rated SAFE, 3 flagged. Scanned against 43 published rules. claude-seo,… - [Claude Skills Managers & CLIs: 191 Security-Graded (2026)](https://agentskillshub.top/best/skill-management-tools/): 191 Claude Skills Managers & CLIs, independently security-graded: 185 rated SAFE, 5 flagged. Scanned against 43 published rules. skills-manager,… - [Open-Source Telegram Bots for Fun & AI: 261 Graded (2026)](https://agentskillshub.top/best/telegram-bot/): 261 Open-Source Telegram Bots for Fun & AI, independently security-graded: 221 rated SAFE, 40 flagged. Scanned against 43 published rules. aiogram,… - [TypeSafe Jev Skills, Agents & MCP Servers: 344 Graded (2026)](https://agentskillshub.top/best/typesafe-jev/): 344 TypeSafe Jev Skills, Agents & MCP Servers, independently security-graded: 333 rated SAFE, 11 flagged. Scanned against 43 published rules.… - [Open-Source Vibe Coding Tools: 30 Security-Graded (2026)](https://agentskillshub.top/best/vibe-coding/): 30 Open-Source Vibe Coding Tools, independently security-graded: 30 rated SAFE. Scanned against 43 published rules. ralphy, vibe-coding-prompt-template,… - [Voice Agents: 30 Security-Graded (2026)](https://agentskillshub.top/best/voice-agent/): 30 voice agents tools, independently security-graded: 29 rated SAFE, 1 flagged. Scanned against 43 published rules. openlive, openclaw-whobot-skill,… - [Web Scraping: 10 Security-Graded (2026)](https://agentskillshub.top/best/web-scraping/): 10 web scraping tools, independently security-graded: 10 rated SAFE. Scanned against 43 published rules. moli, spider, Scrapling. ## Head-to-head comparisons - [autogen vs crewAI: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/autogen-vs-crewai/): Compare autogen and crewAI side by side. Stars, language, license, quality score, and community activity. Find the best autogen vs crewai for your project. - [browser-use vs claude-code: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/browser-use-vs-claude-code/): Compare browser-use and claude-code side by side. Stars, language, license, quality score, and community activity. Find the best browser use vs claude code for… - [claude-code vs skills: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/claude-code-vs-claude-skills/): Compare claude-code and skills side by side. Stars, language, license, quality score, and community activity. Find the best claude code vs claude skills for… - [deepagents vs autogen: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/deepagents-vs-autogen/): Compare deepagents and autogen side by side. Stars, language, license, quality score, and community activity. Find the best deepagents vs autogen for your… - [github-mcp-server vs n8n-mcp: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/github-mcp-vs-n8n-mcp/): Compare github-mcp-server and n8n-mcp side by side. Stars, language, license, quality score, and community activity. Find the best github mcp vs n8n mcp for… - [langchain vs langgraph: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/langchain-vs-langgraph/): Compare langchain and langgraph side by side. Stars, language, license, quality score, and community activity. Find the best langchain vs langgraph for your… - [langgraph vs crewAI: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/langgraph-vs-crewai/): Compare langgraph and crewAI side by side. Stars, language, license, quality score, and community activity. Find the best langgraph vs crewai for your project. - [n8n vs langflow: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/n8n-vs-langflow/): Compare n8n and langflow side by side. Stars, language, license, quality score, and community activity. Find the best n8n vs langflow for your project. - [servers vs awesome-mcp-servers: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/official-mcp-vs-awesome-mcp/): Compare servers and awesome-mcp-servers side by side. Stars, language, license, quality score, and community activity. Find the best mcp servers official vs… - [ollama vs open-webui: Which AI Tool to Choose in 2026?](https://agentskillshub.top/compare/ollama-vs-open-webui/): Compare ollama and open-webui side by side. Stars, language, license, quality score, and community activity. Find the best ollama vs open webui for your… ## Categories - [Best Open-Source MCP Servers (2026) — Security-Graded](https://agentskillshub.top/category/mcp-server/): An MCP server hands an AI agent real capabilities — and often real credentials. Every server here is scored on 10 signals and, once its README has been read… - [Best Claude Skills (2026) — Open-Source, Quality-Scored](https://agentskillshub.top/category/claude-skill/): A Claude Skill runs with your agent's full permissions, so trust matters as much as usefulness. These are the top open-source Claude Skills, each quality-scored… - [Best AI Agent Tools (2026) — Open-Source, Security-Graded](https://agentskillshub.top/category/agent-tool/): Standalone tools that give agents new abilities — automation, retrieval, orchestration, and more. Ranked by a 10-signal composite (not raw stars) and… - [OpenClaw Ecosystem Skills (2026) — Open-Source & Graded](https://agentskillshub.top/category/codex-skill/): Tools built for the OpenClaw and Codex agent runtimes, indexed from GitHub and refreshed every 8 hours. Each is quality-scored, and security-graded once its… - [Best AI Skill — 5917+ Open-Source Skills & MCP Servers](https://agentskillshub.top/category/ai-skill/): Browse 5917+ open-source AI Skill — quality-scored, ranked by stars, compared side-by-side. Updated every 8 hours. - [Best LLM Plugin — 4360+ Open-Source Skills & MCP Servers](https://agentskillshub.top/category/llm-plugin/): Browse 4360+ open-source LLM Plugin — quality-scored, ranked by stars, compared side-by-side. Updated every 8 hours. - [Best YouMind Plugin — 84+ Open-Source Skills & MCP Servers](https://agentskillshub.top/category/youmind-plugin/): Browse 84+ open-source YouMind Plugin — quality-scored, ranked by stars, compared side-by-side. Updated every 8 hours. - [Best Education — 1+ Open-Source Skills & MCP Servers](https://agentskillshub.top/category/education/): Browse 1+ open-source Education — quality-scored, ranked by stars, compared side-by-side. Updated every 8 hours. ## Blog - [How to Check if a Claude Skill Is Safe Before You Install It (2026 Data)](https://agentskillshub.top/blog/check-claude-skill-safety/): A 5-step pre-install check for Claude Skills and MCP servers, backed by November 2026 data: 138,000+ of 276,000+ indexed skills carry rule-based security… - [I Built an 86K AI Skill Directory in 10 Weeks. Here's Why I'm Changing Course.](https://agentskillshub.top/blog/pivot-to-trust-layer/): Agent Skills Hub indexed 86,000 open-source AI skills in 10 weeks. Traffic grew fast. Revenue stayed at $0. Here's what 15 user interviews taught me — and why… - [我用 10 周做了一个 86K 的 AI Skill 目录。为什么现在要转向。](https://agentskillshub.top/blog/pivot-to-trust-layer/zh/): Agent Skills Hub 10 周收录了 86,000 个开源 AI skill。流量涨得飞快,月收入却是 $0。15 个用户访谈告诉我:发现 skill 从来不是问题 —— 这就是 Hub 从「目录」转向「AI agent 部署信任层」的原因。 - [We Security-Graded 117,854 AI Agent Skills. Here's What We Found.](https://agentskillshub.top/blog/securing-117k-ai-skills/): We ran a rule-based security scan over 117,854 indexed AI agent skills and MCP servers. Only 17.7% are popular enough to be graded; among those, 1 in 32 is… - [我们给 117,854 个 AI Agent Skill 做了安全评级,结果如下](https://agentskillshub.top/blog/securing-117k-ai-skills/zh/): 我们用规则引擎扫了 117,854 个收录的 AI agent skill 和 MCP server。只有 17.7% 火到值得评级;在被评级的里面,每 32 个就有 1 个不安全。风险集中在长尾 —— 还有一个 agent 时代独有的新攻击面。 - [How We Secure 43,000+ AI Agent Tools: A Solo Developer's Security Journey](https://agentskillshub.top/blog/securing-43k-ai-tools/): From zero security to 27 detection rules, 5-tier trust hierarchy, and LLM-powered analysis. How one developer built a multi-layered security system for the… - [How I'd Build a SaaS Solo in 6 Weeks Using AI Skills (2026 Stack)](https://agentskillshub.top/blog/skill-stack-solo-saas-2026/): From idea to launch in 6 weeks as a solo founder. The exact AI Skill stack — validation, PRD, multi-agent build, business diagnosis, GTM. With real examples… - [独立开发者 6 周用 AI Skills 做 SaaS(2026 完整 Stack)](https://agentskillshub.top/blog/skill-stack-solo-saas-2026/zh/): 一人公司从想法到上线 6 周。完整 5 层 Skill 栈:验证 → PRD → 多 Agent 团队 → 商业诊断 → GTM。19 个真实项目示范。 ## 中文指南 (scenario guides in Chinese) - [最佳 Blender、Unity、Godot MCP 与 3D Skill AI 工具 (2026)](https://agentskillshub.top/best/3d-mcp/zh/): 让 Claude Code、Codex 等 agent 做 3D 的开源 MCP 服务和 skill,带安全评级:Blender、Unity、Godot、Unreal、CAD、Three.js 和 3D 模型生成。安装前先核查。 - [最佳 AI 设计工具 (2026)](https://agentskillshub.top/best/ai-design/zh/): AI 编程 agent 的设计智能技能——UI 生成、设计系统、原型、稿转码,接 Google Stitch 等。 - [最佳 AI 逆向工程工具 (2026)](https://agentskillshub.top/best/ai-reverse-engineering/zh/): 开源的 AI 逆向工程工具,带安全评级:Ghidra MCP、IDA Pro MCP、移动端的 jadx 与 Frida、逆向 Skill、REA,以及用它重建出来的项目。 - [最佳 去 AI 味 / Humanizer Skill (2026)](https://agentskillshub.top/best/anti-slop/zh/): 开源的去 AI 味与 humanizer skill,带安全评级:让 AI 写的文字、代码和界面读起来像人做的(blader/humanizer、Humanizer-zh、taste-skill),并检测 AI 腔。安装前先核查。 - [最佳 认证与身份 AI 工具 (2026)](https://agentskillshub.top/best/authentication/zh/): 用于认证、OAuth、SSO 与身份管理的 AI agent 工具。 - [最佳 AI 浏览器自动化:browser-use 与 Playwright MCP AI 工具 (2026)](https://agentskillshub.top/best/browser-automation/zh/): browser-use、playwright-mcp、MediaCrawler 等 20+ 开源 AI 浏览器自动化工具对比,带安全评级。 - [最佳 Claude Code Hooks、Subagents 与状态栏 AI 工具 (2026)](https://agentskillshub.top/best/claude-code-hooks/zh/): 开源的 Claude Code hooks、subagents 与 statusline,带安全评级:Hook 守卫与格式化、专家 Agent 合集、用量与上下文状态栏。安装前先核查。 - [最佳 Claude Code 技能 (2026)](https://agentskillshub.top/best/claude-code-skills/zh/): Claude Code skill 合集:一个仓库里装着很多 skill、插件或子 agent,可以一起安装。综合合集、单领域合集、官方与厂商目录,以及最知名的几个 awesome 清单,每个都做了安全评级。 - [最佳 Claude 视频剪辑与 Remotion Skill (2026)](https://agentskillshub.top/best/claude-video-skills/zh/): 在 Claude Code 里剪现成素材、用 Remotion 做新视频的开源 skill 和工具包:HyperFrames、动效、视频剪辑 agent,其中两个项目是用 Claude Opus 5.5 做出来的。这些 skill 不绑定某个模型,其他 Claude / Codex 模型同样能跑。逐个做了安全评级。 - [最佳 命令行工具 (2026)](https://agentskillshub.top/best/cli-tools/zh/): 面向开发者的命令行 AI 工具与终端实用程序。 - [最佳 代码审查 AI 工具 (2026)](https://agentskillshub.top/best/code-review/zh/): 自动捕捉 bug、统一代码风格、提升质量的顶级 AI 代码审查工具。 - [最佳 OpenAI Codex 技能 (2026)](https://agentskillshub.top/best/codex-skills/zh/): OpenAI AI 编程 agent Codex 的最佳技能与插件。 - [最佳 内容分发 AI 工具 (2026)](https://agentskillshub.top/best/content-syndication/zh/): 独立创作者的跨平台内容分发——一次发布到 X、领英、公众号、小红书、B站、Telegram。 - [最佳 内容写作 AI 工具 (2026)](https://agentskillshub.top/best/content-writing/zh/): 用于自动写作、博客生成、文案与文本创作的 AI agent 技能。 - [最佳 AI 生图 Skill:GPT Image、Nano Banana 与 Flux AI 工具 (2026)](https://agentskillshub.top/best/image-generation/zh/): 给 Claude Code、Codex 等 agent 用的开源生图 skill 和 MCP 服务,带安全评级:GPT Image、Nano Banana、Flux、图片编辑和设计素材。安装前先核查。 - [最佳 知识库与 RAG AI 工具 (2026)](https://agentskillshub.top/best/knowledge-base/zh/): 用 RAG 检索增强构建 AI 知识库——导入文档、语义搜索、智能问答。 - [最佳 本地大模型工具 (2026)](https://agentskillshub.top/best/local-llm/zh/): 本地私有化运行 AI 模型、零 API 费用——自托管推理、微调与部署工具。 - [最佳 MCP API 集成 AI 工具 (2026)](https://agentskillshub.top/best/mcp-api/zh/): 把 AI agent 接入外部 API、REST 端点与第三方服务的 MCP 工具。 - [最佳 数据库 MCP:Supabase、Postgres 与 SQL AI 工具 (2026)](https://agentskillshub.top/best/mcp-database/zh/): 让 Claude、Codex 等 AI agent 查询和操作数据库的开源 MCP 服务与工具,带安全评级:Supabase、PostgreSQL、MySQL、SQLite、NoSQL 和分析型数据库。安装前先核查。 - [最佳 GitHub 的 MCP 工具 (2026)](https://agentskillshub.top/best/mcp-for-github/zh/): 接入 GitHub 管理仓库、issue、PR、Actions 与代码审查的 MCP 工具。 - [最佳 Google 服务的 MCP 工具 (2026)](https://agentskillshub.top/best/mcp-for-google/zh/): 把 AI agent 接入 Google Workspace——Gmail、Drive、日历、Sheets 等。 - [最佳 多智能体编排 AI 工具 (2026)](https://agentskillshub.top/best/multi-agent/zh/): 构建与编排多智能体系统的工具,让 AI agent 协作、委派与协调任务。 - [最佳 Obsidian × Claude:AI Skill、MCP 与插件 AI 工具 (2026)](https://agentskillshub.top/best/obsidian-second-brain/zh/): 让 Claude Code、Codex 等 agent 读写和整理 Obsidian 库的开源 skill、MCP 服务和 AI 插件,带安全评级:kepano/obsidian-skills、mcp-obsidian、claude-obsidian、第二大脑工作流。安装前先核查。 - [最佳 Codex 与 Claude Code PPT Skills (2026)](https://agentskillshub.top/best/ppt-presentation/zh/): 生成幻灯片与演示的 AI 技能——HTML deck、麦肯锡风 PPTX、图优先 PPT、营销幻灯。 - [最佳 提示词工程 AI 工具 (2026)](https://agentskillshub.top/best/prompt-engineering/zh/): 面向 LLM 应用的提示词设计、测试、优化与管理工具。 - [最佳 代码重构 AI 工具 (2026)](https://agentskillshub.top/best/refactoring/zh/): 用于自动化代码重构、优化与架构改进的 AI 工具。 - [最佳 密钥检测 AI 工具 (2026)](https://agentskillshub.top/best/secret-detection/zh/): 在密钥、API key、凭证泄露引发安全事故前检测出来的工具。 - [最佳 Claude SEO Skill 与 SEO MCP (2026)](https://agentskillshub.top/best/seo-tools/zh/): 开源的 Claude SEO skill 与 SEO MCP 服务,带安全评级:claude-seo、geo-seo-claude、站点审计与关键词研究 skill、GEO 工具,以及接 Google Search Console、Ahrefs、DataForSEO 的 MCP 服务。安装前先核查。 - [最佳 技能与 Agent 管理 (2026)](https://agentskillshub.top/best/skill-management-tools/zh/): 跨 15+ AI 编程工具组织 Claude/Codex 技能与多智能体工作流的桌面应用、CLI 与会话管理器。 - [最佳 Telegram 机器人 AI 工具 (2026)](https://agentskillshub.top/best/telegram-bot/zh/): 开源 Telegram 机器人框架、MCP 服务器与 AI agent 集成,搭自动化与 AI 助手。 - [最佳 TypeSafe Jev AI 工具 (2026)](https://agentskillshub.top/best/typesafe-jev/zh/): 基于 TypeSafe Jev 决策模型的开源项目,逐个做了安全评级:jev-ultrafast 浏览器 agent、openjev、fast-jev-compaction、Jev 代码审查工具和 MCP 服务器。装之前先查一查。 - [最佳 Vibe Coding 工具栈 AI 工具 (2026)](https://agentskillshub.top/best/vibe-coding/zh/): 独立开发者的编码工具栈——PRD→多智能体构建→发布,vibe coding 全流程模板。 - [最佳 语音智能体 AI 工具 (2026)](https://agentskillshub.top/best/voice-agent/zh/): 实时语音 AI agent——语音进语音出的对话系统,接 Whisper、ElevenLabs、OpenAI Realtime。 - [最佳 网页抓取 AI 工具 (2026)](https://agentskillshub.top/best/web-scraping/zh/): 为网页抓取、数据提取与自动爬取精选最佳 AI Agent 工具与 MCP 服务器。