How are MCP tools accessed within Claude Code?

Tools are automatically prefixed using the format 'mcp__plugin_ _ __ ' and can be viewed using the /mcp command.

Can one Claude plugin use multiple MCP servers?

Yes, you can define multiple servers within a single .mcp.json file or the mcpServers field in your plugin.json to span multiple services.

What MCP server transport types are supported?

This skill supports four primary types: stdio for local processes, SSE for hosted services with OAuth, HTTP for RESTful APIs, and WebSocket for real-time bidirectional communication.

How do I securely handle API keys in my MCP configuration?

It is recommended to use environment variable expansion (e.g., ${MY_API_KEY}) in your configuration files instead of hardcoding sensitive credentials.

MCP Integration

Name: MCP Integration
Author: anthropics

byanthropics

•

291

API Development

Integrates Model Context Protocol (MCP) servers into Claude Code plugins to connect AI workflows with external tools and services.

About

The MCP Integration skill provides comprehensive guidance for developers looking to extend Claude Code's capabilities through the Model Context Protocol. It enables the seamless connection of external services, databases, and APIs to Claude by supporting various transport methods including stdio, SSE, HTTP, and WebSockets. By following this skill's patterns, developers can manage complex OAuth flows, handle environment-specific configurations, and register standardized tools that allow Claude to interact autonomously with third-party platforms and local system resources.

Key Features

Standardized tool naming and lifecycle management for autonomous agentic workflows
Automated OAuth and token-based authentication management for cloud services
291 GitHub stars
Flexible configuration options via dedicated .mcp.json or inline plugin.json files
Support for multiple transport types including stdio, SSE, HTTP, and WebSocket
Dynamic environment variable expansion for secure and portable plugin deployment

Use Cases

Building custom autonomous agents that interact with internal REST APIs through the MCP bridge
Automating tasks across cloud platforms like GitHub, Jira, and Asana via hosted MCP servers
Exposing local databases and file systems as tools for advanced data analysis and code generation

About

Key Features

Standardized tool naming and lifecycle management for autonomous agentic workflows
Automated OAuth and token-based authentication management for cloud services
291 GitHub stars
Flexible configuration options via dedicated .mcp.json or inline plugin.json files
Support for multiple transport types including stdio, SSE, HTTP, and WebSocket
Dynamic environment variable expansion for secure and portable plugin deployment

Use Cases

Building custom autonomous agents that interact with internal REST APIs through the MCP bridge
Automating tasks across cloud platforms like GitHub, Jira, and Asana via hosted MCP servers
Exposing local databases and file systems as tools for advanced data analysis and code generation