Integrations6 min read

Bring skills into every AI tool you use

Install a skill once. Use it in Claude Code, Cursor, Windsurf, Copilot, Cline, or any tool that speaks MCP. No vendor lock-in, no format conversion.

Claude Code

Claude Code has native MCP support. Skills installed via npx @skills-hub-ai/cli install are automatically discovered from ~/.claude/skills/.

Option A: Direct install (no MCP needed)

bash
npx @skills-hub-ai/cli install code-review
# Now type /code-review in Claude Code

Option B: MCP server (all skills as prompts)

bash
claude mcp add skills-hub -- npx @skills-hub-ai/mcp

This registers the MCP server globally. All installed skills appear as prompts across all projects.

Cursor

Cursor supports MCP servers via configuration files. Add the skills-hub server to use your installed skills as prompts.

Step 1: Install skills

bash
npx @skills-hub-ai/cli install code-review --target cursor
# Installs to ~/.cursor/skills/code-review/

Step 2: Configure MCP server

Add to .cursor/mcp.json in your project root (or globally at ~/.cursor/mcp.json):

json
{
  "mcpServers": {
    "skills-hub": {
      "command": "npx",
      "args": ["@skills-hub-ai/mcp"]
    }
  }
}

Step 3: Use skills

Open Cursor, and your installed skills will be available as prompts. The MCP server reads from both ~/.claude/skills/ and ~/.cursor/skills/.

Windsurf

Windsurf (by Codeium) supports MCP servers for extending Cascade with custom tools and prompts. Configure the skills-hub MCP server to bring your skills into Windsurf.

Step 1: Install skills

bash
npx @skills-hub-ai/cli install code-review

Step 2: Configure MCP server

Open Windsurf Settings, navigate to Cascade > MCP Servers, and add a new server. Or add to ~/.codeium/windsurf/mcp_config.json:

json
{
  "mcpServers": {
    "skills-hub": {
      "command": "npx",
      "args": ["@skills-hub-ai/mcp"]
    }
  }
}

Step 3: Restart and use

Restart Windsurf or reload the MCP configuration. Your installed skills will appear as available prompts in Cascade conversations.

GitHub Copilot

GitHub Copilot supports MCP servers in VS Code through the Copilot Chat agent mode. This lets you use skills-hub skills directly in Copilot conversations.

Step 1: Install skills

bash
npx @skills-hub-ai/cli install code-review

Step 2: Configure MCP server

Add to your VS Code settings.json (or workspace .vscode/settings.json):

json
{
  "github.copilot.chat.mcp.servers": {
    "skills-hub": {
      "command": "npx",
      "args": ["@skills-hub-ai/mcp"]
    }
  }
}

Alternatively, add to .vscode/mcp.json in your project:

json
{
  "servers": {
    "skills-hub": {
      "command": "npx",
      "args": ["@skills-hub-ai/mcp"]
    }
  }
}

Step 3: Enable agent mode

In VS Code, open Copilot Chat and switch to Agent mode (click the mode selector at the top of the chat panel). Your skills will be available as MCP prompts.

Cline

Cline (formerly Claude Dev) is a VS Code extension with built-in MCP support. Configure the skills-hub server to use your skills in Cline conversations.

Step 1: Install skills

bash
npx @skills-hub-ai/cli install code-review

Step 2: Configure MCP server

Open Cline settings in VS Code (click the gear icon in the Cline sidebar). Navigate to MCP Servers and add a new server:

json
{
  "mcpServers": {
    "skills-hub": {
      "command": "npx",
      "args": ["@skills-hub-ai/mcp"]
    }
  }
}

Or add the server via Cline's MCP Servers panel > Add Serverbutton, selecting "stdio" as the transport type.

Step 3: Use skills

Once configured, Cline will discover your installed skills as MCP prompts. Reference them in your conversations with Cline to invoke skill instructions.

Any MCP-compatible tool

Any tool that supports the Model Context Protocol can use skills-hub skills. The server uses stdio transport by default.

Generic setup

bash
# Server command:
npx @skills-hub-ai/mcp

# Transport: stdio
# The server scans:
#   ~/.claude/skills/
#   ~/.cursor/skills/
# and serves each SKILL.md as an MCP prompt.

Consult your tool's documentation for how to add an MCP server. Most tools accept a command + args configuration pointing to the server executable.

Next steps