skill4agent MCP Server
# @skill4agent/mcp-server
[δΈζ](https://github.com/osulivan/skill4agent-mcp-server/blob/main/README_CN.md) | English
MCP (Model Context Protocol) Server for skill4agent - Search, view, and install AI skills in AI conversations.
## Features
Provides the following capabilities through the MCP protocol:
- π **Search Skills** - Search AI skills by keyword, with optional category filtering
- π **Get Details** - View complete skill documentation (SKILL.md)
- π¦ **Installation Info** - Get download links and installation commands for skills
## Installation
```bash
npm install -g @skill4agent/mcp-server
```
Or run directly with npx:
```bash
npx @skill4agent/mcp-server
```
## Usage in AI Applications (Claude Desktop, etc.)
Add the following to your Claude Desktop configuration file:
```json
{
"mcpServers": {
"skill4agent": {
"command": "npx",
"args": ["-y", "@skill4agent/mcp-server"]
}
}
}
```
To use a custom API endpoint, add the `env` configuration:
```json
{
"mcpServers": {
"skill4agent": {
"command": "npx",
"args": ["-y", "@skill4agent/mcp-server"],
"env": {
"SKILL4AGENT_API_URL": "https://your-custom-domain.com/api/mcp"
}
}
}
}
```
## Available Tools
### search_skills
Search for AI skills.
**Parameters:**
- `keyword` (required): Search keyword
- `categories` (optional): Category filter, supports both English and Chinese
- `limit` (optional): Limit the number of results, default is 10, maximum is 100
**Example:**
```json
{
"keyword": "React",
"categories": ["Frontend Development"],
"limit": 5
}
```
### get_skill
Get detailed information about a specific skill.
**Parameters:**
- `skillId` (required): Skill ID, obtained from the `search_skills` tool results
**Example:**
```json
{
"skillId": "frontend-design--anthropics-skills"
}
```
### install_skill
Get installation information for a skill.
**Parameters:**
- `skillId` (required): Skill ID
**Example:**
```json
{
"skillId": "frontend-design--anthropics-skills"
}
```
## Environment Variables
| Variable | Description | Default |
|----------|-------------|---------|
| `SKILL4AGENT_API_URL` | skill4agent API endpoint | `https://skill4agent.com/api/mcp` |
## Local Development
```bash
# Install dependencies
npm install
# Build
npm run build
# Development mode (hot reload)
npm run dev
# Run directly
node dist/index.js
```
## Project Structure
```
@skill4agent/mcp-server/
βββ src/
β βββ index.ts # Entry point
β βββ server.ts # MCP Server main logic
β βββ api/
β β βββ client.ts # API client
β βββ tools/
β βββ search.ts # search_skills tool
β βββ detail.ts # get_skill tool
β βββ install.ts # install_skill tool
βββ package.json
βββ tsconfig.json
βββ README.md
```
## Related Links
- [skill4agent](https://skill4agent.com)
- [MCP Documentation](https://modelcontextprotocol.io)
- [NPM Package](https://www.npmjs.com/package/@skill4agent/mcp-server)
## License
MIT
TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose with no overlap: get_skill retrieves detailed information, install_skill provides installation methods, and search_skills finds skills by keyword/category. The descriptions reinforce these distinct roles, making misselection unlikely.
All tools follow a consistent verb_noun pattern with snake_case (get_skill, install_skill, search_skills). The naming is predictable and readable, with verbs that accurately reflect each tool's function without deviation.
With only 3 tools, the set feels thin for a skill management server, as it lacks operations like updating, deleting, or managing skill configurations. While the tools cover basic discovery and installation, the scope suggests more functionality would be expected for a complete skill lifecycle.
The tool surface is significantly incomplete for skill management, missing essential CRUD operations such as create_skill, update_skill, and delete_skill. Agents will hit dead ends when trying to modify or remove skills, limiting the server's utility beyond basic lookup and installation.