MCP Integration FAQs

Question 1

Can I use multiple MCP servers in a single plugin?

Accepted Answer

Yes, you can define multiple servers within your .mcp.json or plugin.json file, allowing your Claude Code plugin to orchestrate workflows across various services simultaneously.

Question 2

What is the Model Context Protocol (MCP)?

Accepted Answer

MCP is an open standard that enables AI models to connect to external data sources and tools, providing a structured way for Claude Code to interact with APIs, databases, and local file systems.

Question 3

How do I secure API keys in my MCP configuration?

Accepted Answer

You should use environment variable expansion (e.g., ${API_KEY}) in your configuration files rather than hardcoding credentials. This ensures keys are pulled from the user's shell environment and not committed to version control.

Question 4

What is the difference between stdio and SSE server types?

Accepted Answer

stdio executes a local process as a child of Claude Code (ideal for local utilities), while SSE (Server-Sent Events) connects to remote, hosted services over the web and often handles automated OAuth flows.

Question 5

How do I reference an MCP tool in a command?

Accepted Answer

Use the specific tool name format 'mcp__plugin____' and ensure the tool is explicitly listed in the 'allowed-tools' section of your command's frontmatter.

MCP Integration

MCP Integration

主要功能

使用场景

主要功能

使用场景