Setup mcp

Skill az9713/youtube-transcript-to-skills/.claude/skills/setup-mcp

Extract Claude Code skills from AI workflow YouTube transcripts. 10 ready-to-use skills + a meta-skill that automates the extraction process. Built from John Kim's 50 Claude Code tips.

Install

npx -y skills add az9713/youtube-transcript-to-skills --skill setup-mcp

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Find, install, and configure an MCP server for a specific service. Handles discovery, installation, and verification.

SKILL.md

3.2 KB, 780 tokens by cl100k_base, as published. Nobody here has run it

Setup MCP Server

Find, install, and configure an MCP (Model Context Protocol) server for a specific external service.

Usage

/setup-mcp <service-name>

Examples:

/setup-mcp github
/setup-mcp postgres
/setup-mcp figma
/setup-mcp slack

Procedure

Step 1: Identify the Service

Parse $ARGUMENTS to determine what service the user wants to connect. Common services:

github — GitHub API (issues, PRs, repos)
postgres / mysql / sqlite — Database access
figma — Design file access
slack — Messaging
linear — Project management
notion — Documentation
filesystem — Local file access with sandboxing
brave-search — Web search
memory — Persistent knowledge graph

Step 2: Search for MCP Server

Search for an appropriate MCP server:

Check the official MCP servers list by searching the web for: site:github.com modelcontextprotocol/servers [service-name]
Also search for community servers: mcp server [service-name]
Check npm: npm search mcp-server-[service-name]

Evaluate candidates on:

Stars / downloads (popularity = maintenance likelihood)
Last updated (stale = risky)
Official vs community (prefer official)

Step 3: Present Options

Show the user what was found:

Found MCP servers for [service]:

1. [name] (official) — [description]
   Stars: [N] | Last updated: [date]
   Install: [command]

2. [name] (community) — [description]
   Stars: [N] | Last updated: [date]
   Install: [command]

Which would you like to install?

Step 4: Install

Based on the server type, install appropriately:

For npx-based servers (most common): Add to .claude/settings.json under mcpServers:

{
  "mcpServers": {
    "[service-name]": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-[name]"],
      "env": {
        "API_KEY": "your-api-key-here"
      }
    }
  }
}

For Docker-based servers:

{
  "mcpServers": {
    "[service-name]": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "[image-name]"]
    }
  }
}

Step 5: Configure Authentication

If the MCP server requires API keys or tokens:

Tell the user what credentials are needed
Ask them to provide the credentials
Never store credentials directly in settings.json — use environment variables
Show how to set the env var: export MCP_[SERVICE]_KEY=your-key

Step 6: Verify Installation

Tell the user to run /mcp to check the server status
The server should appear in the list with a "connected" status
If it fails, check:
- Is the command/path correct?
- Are environment variables set?
- Are network/firewall rules blocking it?

Step 7: Warn About Token Costs

Important caveat to share with the user:

"Note: MCP servers add tools to Claude's context, which increases token usage. Each MCP server's tool descriptions are loaded into every message. If you notice higher token consumption, consider disabling MCP servers you're not actively using."