Claude MCP Guide: Connect Claude Code to Tools and Data
Model Context Protocol standardizes how AI applications connect to external tools and data. In Claude Code, MCP can expose issue trackers, databases, monitoring systems, SaaS applications and internal APIs.
Quick answer
Use MCP when Claude needs capabilities or information outside the repository. Current Claude Code docs recommend HTTP for remote servers, stdio for local processes, and describe SSE as deprecated where HTTP is available.
Remote HTTP servers
HTTP is the recommended remote transport in current Claude Code documentation and supports hosted integrations and authentication patterns.
Local stdio servers
A stdio MCP server runs as a local process and is a natural fit for scripts or tools that need local system access.
Project-scoped configuration
Claude Code can share MCP server definitions through .mcp.json using project scope. Review trust and permissions because cloned repositories cannot silently approve their own project MCP servers.
Project .mcp.json example
{
"mcpServers": {
"shared-server": {
"type": "http",
"url": "https://mcp.example.com/mcp"
}
}
}Official Claude documentation
Claude Code changes quickly. Use Anthropic’s documentation as the source of truth for current configuration fields and behavior.
Open Anthropic docs →Related Claude tools and guides
Frequently asked questions
What is MCP?
MCP is an open protocol for connecting AI applications to tools and data sources.
Which remote MCP transport should I use?
Use HTTP when available. Claude Code documentation recommends HTTP for remote servers and labels SSE as deprecated.
What is .mcp.json?
It is the project-scoped MCP configuration file used for shared server definitions.