Skip to content
Get packs

MCP Integration Guide

Teaches the agent to bundle and configure MCP servers in Claude Code plugins

Source Download skill
Repository
anthropics/claude-plugins-official
Stars
45,770
Licence
Apache-2.0
Files
7 files

What it does

This skill gives an AI agent detailed guidance for connecting Claude Code plugins to outside services through the Model Context Protocol (MCP), a standard way to give an AI access to external tools and data. Once installed, the agent knows how to write a .mcp.json or plugin.json config for local (stdio), hosted (SSE, HTTP) and WebSocket servers, how to pass environment variables and tokens, how MCP tools get named, and how to pre-allow specific tools in commands. It is documentation and templates; it does not itself connect to any service.

Great for

  • Plugin developers who want to bundle an external service (such as Asana or a database) with a Claude Code plugin
  • Developers setting up their first MCP server configuration and unsure which transport type to choose
  • Teams writing commands or agents that need to call MCP tools with tight tool permissions

Skip it if

  • People who just want to use an existing MCP server without building a plugin; they can configure it directly in Claude Code settings
  • Configurations that paste real API tokens into config files, since the guidance can encourage committing secrets to version control if handled carelessly
  • Security-sensitive setups that rely on wildcard tool permissions, which the guide itself says to avoid

Try asking

  • “Add an MCP server for my Asana account to this plugin and configure it in .mcp.json”
  • “Should I use stdio or SSE for connecting my plugin to a hosted service?”
  • “Set up a token-authenticated HTTP MCP server with an API key from an environment variable”
  • “Show me how to pre-allow only specific MCP tools in my slash command's frontmatter”
Step-by-step guide: how to use MCP Integration GuideInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 7 original files, the Apache-2.0 licence and an attribution note.

    mcp-integration.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip mcp-integration.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip mcp-integration.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Use ${CLAUDE_PLUGIN_ROOT} for paths so the plugin stays portable, and reference secrets as environment variables rather than literal values
  • Pre-allow specific MCP tool names in allowed-tools instead of wildcards to limit what the agent can do
  • Restart Claude Code after changing MCP configuration, then check the /mcp command to confirm servers loaded

What's inside

  • examples/http-server.json
  • examples/sse-server.json
  • examples/stdio-server.json
  • references/authentication.md
  • references/server-types.md
  • references/tool-usage.md
  • SKILL.md

Redistributed unmodified from anthropics/claude-plugins-official under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.