Compare skills by installs, and MCP servers and marketplaces by GitHub stars. Skillselion Research publishes dated studies of the catalog with open datasets, free to cite.
For agentsclaude mcp add skillselion -- npx -y skillselion-mcp63,961 skills indexed
Last updated October 22, 2026
Quickly generate short, design-led motion graphics such as kinetic typography, animated data visualizations, logo stings, lower-thirds, or social overlays.
Force the simplest, shortest, most minimal solution at every step and eliminate unnecessary code, dependencies, and features.
Performs a manual six-phase security audit of a codebase to find exploitable vulnerabilities with real impact, each backed by a concrete attack scenario.
Remove repetitive, bloated, and over-abstracted patterns from AI-generated code while preserving original behavior.
Override generic LLM frontend output with intentional, metric-driven UI/UX taste.
Run consistent multi-axis code reviews on changes produced by themselves, other agents, or humans before merging.
Instantly recall decisions, fixes, and patterns from past Claude sessions without repeating work.
So their agent can intelligently recommend and install the right modular skills from the open ecosystem when they ask how to do something.
Have an agent relentlessly pressure-test their plan or design until every decision branch is resolved and shared understanding is reached.
Give their coding agent reliable, high-fidelity control over any website or Electron desktop app.
Pressure-test architecture decisions against their live codebase and living documentation before writing production code.
Scaffold, lint, inspect, preview, render, and troubleshoot HyperFrames video compositions from the terminal.
Safely deepen clusters of shallow modules into cohesive, testable units while respecting their external dependencies.
Generate complete HTML-based video compositions with synchronized animations, captions, voiceovers, and scene transitions using a single agent prompt.
Generate distinctive, production-grade frontend interfaces and components that avoid generic AI aesthetics.
Automatically apply Vercel’s official React and Next.js performance rules during code generation and review.
Automatically audit frontend code against the latest Vercel Web Interface Guidelines for consistency, accessibility, and UX quality.
Override generic LLM frontend output with intentional, metric-driven UI/UX taste.
Make every AI-generated interface look and feel like it was crafted by a $150k premium design agency.
Upgrade an existing website or web app from generic AI-generated styling to premium, intentional design while preserving the current tech stack.
Generate clean, editorial-style web interfaces with warm monochrome palettes, typographic hierarchy, bento grids and flat components that feel premium and utilita
superpowers-dev is Obra's Claude Code plugin marketplace that publishes the Superpowers core skills library as one installable plugin.
ECC is a marketplace that installs a single harness-native operator plugin bundling agents, skills, hooks, rules, and reusable workflows for coding agents.
anthropic-agent-skills is Anthropic's Claude Code plugin marketplace that lists three plugins: a document-processing suite (xlsx, docx, pptx, pdf), a broad example-skills collection, and a Claude API documentation skill.
prompts.chat is a Claude Code plugin marketplace from Fatih Kadir Akin, creator of the prompts.chat prompt collection (the repo behind Awesome ChatGPT Prompts).
Ponytail marketplace plugin adds lightweight project-management workflow support to Claude Code, helping bundle tasks, track multi-step build progress, and keep agent implementation sessions organized across longer feature work.
A Claude Code plugin marketplace exposing one plugin from the ui-ux-pro-max-skill repository.
open-design is nexu-io's Claude Code marketplace that installs one Open Design plugin, a stdio MCP server backed by a local `od` daemon.
Sequential Thinking is one of the official Model Context Protocol reference servers.
Trending hip-hop artist momentum scores across four cultural dimensions.
Web scraping with stealth HTTP, real browsers, and Cloudflare bypass.
Context7 is Upstash's Model Context Protocol server for live library documentation.
MCP server for Chrome DevTools
Lets AI clients search, explore, query, and visualize data in a Metabase instance.
Persistent codebase knowledge graph.
Skillselion indexes the agent ecosystem and runs ongoing research on it: dated censuses of security, copying, usage and maintenance, with the raw dataset behind every number. Free to cite - browse Skillselion Research. How the numbers are made: methodology.
Market stat of the week0 of 8,433 catalogued MCP servers report an install count, so nobody can rank them by real usageMCP Server Census →
Hands-on guides to the most-installed agent skills - what each one does and when to use it - browse all guides.
The core toolkit for anything agentic - agent building, LLM integration, RAG and generative media. Skills and MCP servers that give your coding agent new capabilities.
The visual layer - UI and UX design, CSS and Tailwind, design systems, animation, accessibility and design-to-code. For shipping a real frontend with Claude Code, Codex or Cursor.
The server side - APIs, databases, full-stack scaffolding, data analytics and ML. Tools for wiring up the backend and data layer of a real product.
The everyday inner loop - CLI and terminal, Git and pull requests, debugging and per-language helpers (Python, Go, Rust and more) that make coding faster.
Take the repetitive work off your plate - CI/CD, release automation, browser and web automation, and DevOps workflows your agent can drive end to end.
Keep quality high before you ship - automated testing, QA and code review skills that catch bugs and enforce standards your agent can run on every change.
Ship safer - security and pentesting tools that find vulnerabilities, harden your stack and review code for risks before they reach production.
Run what you shipped - observability, monitoring and cloud infrastructure tools, from logs and metrics to managing the cloud your product runs on.
Stay organized while you build - documentation, planning, research, productivity and communication skills that scope the work and write it up.
An automatically updated directory of skills, MCP servers and marketplaces for Claude Code, Codex, Cursor and other coding agents. Skills are ranked by installs; MCP servers and marketplaces are ranked by GitHub stars. Skillselion Research also publishes dated censuses and open datasets measuring the ecosystem.
Yes - that is what separates it from a plain directory. Skillselion Research runs ongoing censuses of the agent ecosystem (security audit coverage, verified copies, install-weighted use cases, code maintenance) plus an always-current statistics reference. Every edition is frozen on publication, free to cite with attribution, and ships its raw dataset as CSV. How the numbers are made is at /research/methodology.
The catalog updates automatically. Skills are ranked by installs reported by the skills.sh registry; MCP servers and marketplaces are ranked by GitHub stars because they do not report comparable install counts.
No. Browsing, search and the brief are free.
No. Independent project, not affiliated with or endorsed by any tool maker.
No - everything here is free to browse without one.
Not ourselves. Listings surface third-party audit verdicts (Socket, Snyk and other skills.sh auditors) where they exist - outside signals we pass through, not a scan we run and not a safety guarantee. Our Agent Skill Security Census measures that coverage across the whole ecosystem: 87.8% of listings have never been audited by anyone. Review a tool before you install it.
Send us a note through the Feedback page. We read every message - whether it’s a stale entry to fix or a tool we should add. The directory updates automatically, with the ranking method explained on each catalog.
This week in AI coding
Five minutes, every Monday - the tools, releases and tactics for developers.
