skills-master-mcp
Connects AI coding agents to the SkillsMP marketplace, allowing users to search, read, and install over 8,000 community-made skills. It enables agents to gain new capabilities either through on-the-spot instruction or permanent installation without requiring an API key.
README
<h1 align="center">skills-master-mcp</h1>
<p align="center"> <a href="https://www.npmjs.com/package/skills-master-mcp"><img src="https://img.shields.io/npm/v/skills-master-mcp?color=blue&label=npm" alt="npm version" /></a> <a href="https://www.npmjs.com/package/skills-master-mcp"><img src="https://img.shields.io/npm/dm/skills-master-mcp?color=green" alt="npm downloads" /></a> <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/license-MIT-yellow" alt="License: MIT" /></a> <a href="https://nodejs.org"><img src="https://img.shields.io/badge/node-%3E%3D18-brightgreen" alt="Node.js" /></a> <a href="https://modelcontextprotocol.io"><img src="https://img.shields.io/badge/MCP-compatible-purple" alt="MCP Compatible" /></a> </p>
<p align="center"> <a href="#quick-start">Quick Start</a> • <a href="#example">Example</a> • <a href="#tools">Tools</a> • <a href="#how-it-works">Architecture</a> • <a href="#supported-agents">Agents</a> </p>
What is this?
A Plug & Play MCP server that connects your AI coding agent to the SkillsMP marketplace - 8,000+ community-made skills.
You can search for skills, easily install them (to ~/.claude/skills/) or read them directly into your agent's context (without installing).
No API key needed.
One command to set up.
The key idea is to help lazy people like me to use skills more often, and save time and tokens for our dear agents.
Example
Here’s a real use case: converting a Markdown file to PDF.
You can either have the agent learn how to do it via web search, or simply import the skill.
Super efficient and fast.
You: "I need to convert README.md to a PDF"
Agent: searches SkillsMP for "markdown to pdf"
→ finds a skill for it
→ reads the SKILL.md content from GitHub
→ now has the full instructions in context
Agent: "I found a skill for this. It uses Puppeteer to render
the markdown and save it as PDF. Let me do that now."
...converts your file using the skill's instructions.
You: ":))))"
The skill was never installed to your ~/.claude/skills
The agent just read it, learned the approach, and executed it.
One-shot use.
If you want a skill permanently, you can install it too:
You: "Install that markdown-to-pdf skill for Claude Code"
Agent: runs install_skill
→ skill is now saved to .claude/skills/
→ available in every future conversation
Quick Start
No API key
No .env
No configuration
Just install and run
Claude Code
claude mcp add skills-master -- npx skills-master-mcp
Cursor
Add to .cursor/mcp.json:
{
"mcpServers": {
"skills-master": {
"command": "npx",
"args": ["skills-master-mcp"]
}
}
}
Any MCP-Compatible Client
npx skills-master-mcp
Tools
| Tool | Input | Output |
|---|---|---|
search |
keywords (e.g., "react testing") |
List of matching skills with name, author, stars, links |
ai_search |
plain English (e.g., "how to build REST APIs") |
Semantically relevant skills ranked by relevance |
read_skill |
GitHub owner + repo + path | The skill's full SKILL.md content, loaded into agent context |
install_skill |
GitHub source + skill names + target agents | Skill permanently saved to agent's skills directory |
remove_skill |
skill names | Deletes the skill files from agent's skills directory |
<details> <summary>Full parameter reference</summary>
search
| Parameter | Type | Default | Description |
|---|---|---|---|
query |
string | required | Search terms |
page |
number | 1 |
Page number |
limit |
number | 20 |
Results per page (max: 100) |
sort_by |
string | "stars" |
"stars" or "recent" |
response_format |
string | "markdown" |
"markdown" or "json" |
ai_search
| Parameter | Type | Default | Description |
|---|---|---|---|
query |
string | required | Natural language query |
response_format |
string | "markdown" |
"markdown" or "json" |
read_skill
| Parameter | Type | Default | Description |
|---|---|---|---|
owner |
string | required | GitHub user/org |
repo |
string | required | Repository name |
path |
string | — | Path to skill folder |
branch |
string | "main" |
Git branch |
install_skill
| Parameter | Type | Default | Description |
|---|---|---|---|
source |
string | required | GitHub owner/repo or full URL |
skills |
string | required | Skill names (comma-separated) |
agents |
string | required | Target agents (comma-separated) |
global |
boolean | false |
Install globally (user-level) |
remove_skill
| Parameter | Type | Default | Description |
|---|---|---|---|
skills |
string | required | Skill names to remove (comma-separated) |
agent |
string | "claude-code" |
Target agent |
global |
boolean | false |
Remove from global scope |
</details>
Supported Agents
| Agent | Install | Remove |
|---|---|---|
| Claude Code | Yes | Yes |
| Cursor | Yes | Yes |
| Codex | Yes | — |
| OpenCode | Yes | — |
| GitHub Copilot | Yes | — |
| Roo | Yes | — |
| Antigravity | Yes | — |
How It Works
Your AI Agent
│
▼
skills-master-mcp (local, via npx)
├── search / ai_search ──▶ Proxy Backend ──▶ SkillsMP API
├── read_skill ──────────▶ GitHub (direct)
├── install_skill ───────▶ npx add-skill
└── remove_skill ────────▶ Local file deletion
Search requests go through a proxy backend so you don't need an API key. Everything else is direct.
Configuration
Nothing to configure. Optional env vars for advanced use:
| Variable | Default | Description |
|---|---|---|
SKILLS_MASTER_BACKEND_URL |
Production URL | Override backend endpoint |
TRANSPORT |
stdio |
stdio or http |
PORT |
3000 |
HTTP port (when TRANSPORT=http) |
Links
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。