mcp-server-installer

MCP server configuration adds Model Context Protocol servers to Claude Code at the user level in ~/.claude/mcp.json. This Skill automates the process by collecting server details (name, transport type, and transport-specific parameters), prompting for environment variables, and safely merging the new server into your existing configuration while preserving formatting.

MCP server setup supports three transport types: stdio (command-based), HTTP (URL-based), and SSE (Server-Sent Events). Each transport requires different configuration details—stdio needs the command and arguments, while HTTP and SSE need the server URL and optional headers. The configuration process guides you through the correct fields for your chosen transport.

Yes, the configuration process documents both required and optional environment variables for your MCP server. It prompts you for each variable, validates them, and includes guidance in the configuration output so you know which variables the server needs and how to set them before restarting Claude Code.

Yes, the configuration merge reads your existing mcp.json, validates the JSON structure, and safely adds the new server without overwriting or corrupting existing entries. It preserves your current formatting and settings while integrating the new MCP server configuration.

You can run the configuration process multiple times to add different MCP servers. Each run safely merges a new server entry into your ~/.claude/mcp.json, so you can build up a collection of stdio, HTTP, and SSE servers without losing previous configurations.

After configuration completes, the Skill produces a testable MCP server entry with all transport details, environment variable documentation, and restart steps included. The output guides you on required environment variable setup and when to restart Claude Code for the new server to take effect.