MCP Integration for Claude Code FAQs

Question 1

Can I bundle multiple MCP servers in a single Claude Code plugin?

Accepted Answer

Yes, you can configure multiple servers by using a dedicated .mcp.json file at the plugin root or by adding multiple entries to the mcpServers field in your plugin.json.

Question 2

What transport protocols does the MCP Integration skill support?

Accepted Answer

It supports stdio for local processes, SSE (Server-Sent Events) for cloud services, HTTP for REST APIs, and WebSockets for real-time bidirectional communication.

Question 3

How do I reference MCP tools within my commands safely?

Accepted Answer

Tools are automatically prefixed using a specific format. For security, you should explicitly list the required tool names in the 'allowed-tools' section of your command's frontmatter rather than using wildcards.

Question 4

Is it secure to use API keys in the MCP configuration?

Accepted Answer

Yes, as long as you use environment variable expansion (e.g., ${API_KEY}) to reference secrets stored in your shell environment instead of hardcoding them in your JSON files.

Question 5

How does this skill handle authentication for external services?

Accepted Answer

The skill provides patterns for managed OAuth flows in SSE/HTTP servers and supports token-based authentication via environment variables and custom headers.

MCP Integration for Claude Code

About

Key Features

Use Cases

MCP Integration for Claude Code

About

Key Features

Use Cases