AI Agent Skills Hub
1,220 MCP skills, workflows, and integrations with source links and installation guidance.
Review labels and stored grades are catalog metadata, not proof of a completed security audit or a guarantee of safety. Read the review methods and limitations.
Direct answer
AgentSkillsHub is an MCP server and AI agent skills directory with source-based editorial guidance.
Use it when you need to find, compare, and install agent skills without relying on package names alone. The site turns public repo signals, permission context, install guidance, and editorial review into searchable pages that AI coding teams can use before they add a tool to a local or production workflow.
Best use
Shortlist MCP servers, Claude Code skills, and agent workflow extensions before install.
Decision signals
Security grade, permission scope, source evidence, install path, and workflow fit.
Main routes
Search the skill directory, compare MCP guides, then run repo preflight in the analyzer.
Fast answers for AI search
Agent skills hub, Cursor skills, MCP agent skills, and reusable skill libraries
Recent AI citations over-indexed on the homepage. These focused guides give search engines and AI answer systems a better citation target for the exact question instead of forcing every query through one page.
Agent Skills Hub
Definition, safety checklist, and the fastest path into the skills directory.
Open guideCursor Skills Hub
MCP servers, rules, and workflow skills that fit Cursor without overloading the editor.
Open guideMCP Agent Skills
How MCP servers become repeatable agent workflows with clear safety boundaries.
Open guideSkills Library for Agents
A maintainable structure for internal skill libraries, categories, and versioning.
Open guideMarket pulse
Agent skills are moving too fast for a static directory
Explore trending skills, verify MCP server sources, and follow setup guides for your workflow. Check maintenance activity and requested permissions before installing a new tool.
May 18-24
Weekly trend pulse
Explore MCP registries, workflow automation, browser agents, memory systems, local runtimes and research agents.
Source-first
Verified MCP discovery
Use official, registry, and app-source links before trusting a package name from a fast-refresh directory.
Browser workflows
Browser automation lane
Browser-use and Stagehand-style workflows need screenshots, replay, credential boundaries, and rollout controls.
OpenClaw // Featured
OpenClaw Skills Leaderboard
Ranked by a catalog formula combining recorded adoption and grade metadata from 2,993 cleaned OpenClaw skills.
Top OpenClaw Skills
Top 100self-improvement
BCaptures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.
capability-evolver
BA self-evolution engine for AI agents. Analyzes runtime history to identify improvements and applies protocol-constrained evolution.
byterover
BManages project knowledge using ByteRover context tree. Provides two operations: query (retrieve knowledge) and curate (store knowledge). Invoke when user requests information lookup, pattern discovery, or knowledge persistence. Developed by ByteRover Inc. (https://byterover.dev/)
Agent Browser
BA fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured commands.
proactive-agent
BTransform AI agents from task-followers into proactive partners that anticipate needs and continuously improve. Now with WAL Protocol, Working Buffer, Autonomous Crons, and battle-tested patterns. Part of the Hal Stack 🦞
clawddocs
BClawdbot documentation expert with decision tree navigation, search scripts, doc fetching, version tracking, and config snippets for all Clawdbot features
bird
BX/Twitter CLI for reading, searching, and posting via cookies or Sweetistics.
auto-updater
BAutomatically update Clawdbot and all installed skills once daily. Runs via cron, checks for updates, applies them, and messages the user with a summary of what changed.
What is OpenClaw?
OpenClaw is an open skill ecosystem for agent workflows. AgentSkillsHub tracks its most useful skills with structured metadata and security-first scoring.
Learn OpenClawRecorded Grades (A/B)
2,555 of 2,993 OpenClaw entries (85%) have stored A/B grades. This is a different dataset from the main directory, not a safety pass rate.
OpenClaw Tutorials
Top Agent Skills Leaderboard
One entry per repository, ordered by stored repository stars
Hugging Face Transformers is the standard model framework for teams that need local inference, multimodal model experiments, and repeatable agent evaluation workflows without locking every prototype into one hosted API.
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
An agentic skills framework & software development methodology that works.
Reference / test server with prompts, resources, and tools
OpenAI Cookbook is the practical example library many teams use to turn API docs into working patterns for RAG, embeddings, function calling, evals, and agent workflows.
mem0 gives AI agents a memory layer for retaining user preferences, facts, and cross-session context without forcing every conversation into a single prompt.
LiteLLM gives agent teams a provider-neutral gateway for routing OpenAI-style calls across Anthropic, Azure OpenAI, Bedrock, Vertex AI, OpenAI, and many other model providers.
ToolJet is the open-source foundation of ToolJet AI - the enterprise app generation platform for building internal tools, dashboard, business applications, workflows and AI agents
A curated list of awesome skills, hooks, slash-commands, agent orchestrators, applications, and plugins for Claude Code by Anthropic
Installable GitHub library of 1,370+ agentic skills for Claude Code, Cursor, Codex CLI, Gemini CLI, Antigravity, and more. Includes installer CLI, bundles, workflows, and official/community skill coll
Built by Agent Skills Hub
Original workflows, with real output
Start with static release preflight: a downloadable Skill, local Python checker, practice fixture and a real replay of our six-guide release. Free, open source and explicit about what was tested.
Explore original workflows →Product updates · checked September 12, 2026
New in agents and plugins
Explore the official announcements. These are external products, not integrations provided by SkillsHub.
Open-source delivery workflow
Mastra Factory
Connect GitHub issues to persistent coding agents for triage, planning, implementation and PR review, with human checkpoints.
Read the practical guideRole-based business agents
Salesforce Agentforce
Salesforce’s new role-based agents include Casey for customer support, now generally available; see the announcement for each agent’s availability.
Read the Salesforce announcementPlugin evaluation · 2.1.269
Claude Code plugin eval
Run a plugin’s evaluation suite with claude plugin eval and review scored JSON and HTML reports, as documented in the official changelog.
Read the Claude Code changelogTrending Picks This Week
Selected entries from the directory
Hugging Face Transformers
Hugging Face Transformers is the standard model framework for teams that need local inference, multimodal model experiments, and repeatable agent evaluation workflows without locking every prototype into one hosted API.
Everything Claude Code
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Superpowers
An agentic skills framework & software development methodology that works.
Top Agent Skills Collections
Curated skill bundles for common workflows
E-commerce Automation
Shopify, Stripe, inventory management skills
Developer Toolkit
GitHub, CI/CD, code review, deployment skills
Marketing & SEO Suite
Content, analytics, social media, SEO skills
Finance & Operations
Invoicing, expense tracking, payment processing
Workflow Automation
Zapier-style orchestration, data transforms
Recorded Grade A
Recorded labels, not security certifications
Source context before installation
Agent Skills Hub brings together repository links, installation notes, and editorial guidance. Review coverage varies by listing; a stored grade alone does not establish what code was checked.
Use the registry to compare repo signals, validate security posture, and track updates over time. The goal is simple: ship faster while keeping your workflow trustworthy.
What to check before installing
- Hardcoded secrets and token leaks
- Unsafe filesystem access and path traversal
- Unpinned dependencies and suspicious install scripts
- Network calls to unknown endpoints
Showcase // 01
Featured Solution Blueprints
E-commerce Automation
Automatically discover trending items, generate SEO descriptions, and upload to Shopify in minutes.
Smart Expense Audit
Extract invoice data via OCR, verify against policy, and auto-sync to QuickBooks or Google Sheets.
AI Content Engine
Crawl keywords, analyze competitor Gaps, and publish viral threads to X based on trending topics.
Showcase // 02
Official Integrations
Everything Claude Code
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Superpowers
An agentic skills framework & software development methodology that works.
Everything
Reference / test server with prompts, resources, and tools
Showcase // 03
Community Favorites
21st.dev Magic
Generate editable frontend UI components from natural-language requests inside supported AI coding clients.
Geo Seo Claude
GEO-first SEO skill for Claude Code. Comprehensive AI search optimization for any website citability scoring, AI crawler analysis, brand authority, schema markup, platform-specific optimization, and
Mcp Atlassian
MCP server for Atlassian tools (Confluence, Jira)
Star counts are stored repository snapshots, not live measurements or per-skill ratings. Listings from the same repository can share a count. Both repository leaderboards on this page show each repository once. How to interpret repository stars.
Showcase // 04
Skills Leaderboard
Industry Segments
Sponsorship & Partners
Reach 10,000+ AI developers and technical decision makers.
Apply for PlacementEmail: [email protected]
Threat Intelligence
Why Security Scoring Matters?
Grades can be imported defaults or metadata-derived labels. They do not establish a vulnerability rate or prove that every repository was scanned. Read the source, requested permissions, and dated evidence for the specific version you plan to use.
Arbitrary code execution
Skills using eval() or exec() can run malicious Python code on your local machine instantly.
API Key leakage
Hardcoded credentials or improper environment handling can silently exfiltrate your OpenAI/Claude API keys to third-party endpoints.
Detection Logic →Unbounded file access
Unbounded fs.read allows skills to steal SSH keys, .env secrets, or browse history without prompting.
What Are AI Agent Skills?
AI Agent Skills are modular, reusable capabilities that extend the functionality of AI agents like Claude, ChatGPT, and other LLM-powered assistants. Think of them as plugins or extensions that give your AI agent superpowers — from accessing external APIs to executing complex workflows.
With the rise of Model Context Protocol (MCP) and similar standards, developers can now build and share skills that work across multiple AI platforms. This creates an ecosystem where one skill can be used by thousands of developers, dramatically accelerating AI agent development.
Why Security Matters for Agent Skills
Unlike traditional software libraries, agent skills often have broad system access — they can read files, make network requests, and execute code. A malicious or poorly-written skill could expose sensitive data, leak API keys, or compromise your entire development environment.
Agent Skills Hub helps you find source repositories and understand installation requirements. Check shell execution, credential handling, filesystem scope, and remote requests before running a third-party skill. Directory inclusion does not certify its safety.
Popular Use Cases
- Development Tools: GitHub integration, code review automation, deployment pipelines
- Finance & Ops: Stripe payment processing, invoice generation, expense tracking
- E-commerce: Shopify store management, inventory updates, product SEO
- Marketing: Social media scheduling, SEO analysis, content generation
- Workflows: Zapier-style automation, data transformation, API orchestration
Getting Started
Browse 1,220 directory entries by name, category, or platform. Third-party tools may require paid accounts or API usage. AgentSkillsHub also offers commercial review services; directory grades and review labels are not independent certifications or proof that a paid review occurred.
Mission-Critical
Agent Skills
Get the "VERIFIED SECURE" internal node badge.
Boost user trust and get priority routing in our directory.
Includes manual static analysis & security certificate
Frequently Asked Questions
What is AgentSkillsHub?
AgentSkillsHub is a curated directory of MCP servers and AI agent skills for developers building with Claude Code and similar AI coding assistants. Each skill listing includes installation guidance, security scoring, and compatibility context so teams can evaluate and adopt agent tools with confidence.
What is an MCP server?
An MCP (Model Context Protocol) server is a lightweight integration that gives AI coding assistants like Claude Code structured access to external tools and data. MCP servers expose operations — such as reading files, querying databases, or calling APIs — as typed tools that the AI agent can invoke during a coding session without leaving the terminal.
How do I install an MCP skill in Claude Code?
Use the command `claude mcp add <skill-name> -- npx <package-name>` in your terminal. Each skill page on AgentSkillsHub includes the exact install command, required environment variables, and a quick-start guide so you can go from zero to running in under five minutes.
Are all MCP servers safe to use?
No. Review labels and stored grades are screening metadata, not safety guarantees. Some labels originate from import defaults. Review the source, permissions, version, and dated findings before installation.
What is the difference between an MCP server and an agent framework?
MCP servers handle specific integrations — connecting your AI assistant to a single tool like GitHub, Airtable, or Stripe. Agent frameworks (like LangGraph, CrewAI, or AutoGen) orchestrate multi-step reasoning and tool use across many integrations. Most production agent pipelines combine a framework for orchestration with multiple MCP servers for individual tool access.
How often is the AgentSkillsHub directory updated?
Updates vary by listing. A catalog refresh timestamp does not establish that source code was rescanned. Check the linked upstream project and any dated review evidence for the current version.
