liveBook · Manning

Claude Desktop

As explained in AI Agents in Action, Second Edition

Claude Desktop is a desktop application that provides access to Claude models and supports local and remote Model Context Protocol server connections.

Claude Desktop is Anthropic’s free desktop application for accessing Claude and connecting local or remote Model Context Protocol (MCP) servers. MCP gives Claude access to tools, prompts, and resources through a client configuration. section 3.2

A practical Claude Desktop integration follows this path:

  • Create an MCP server, such as a Python server exposing a research-sources tool. section 3.2.1
  • Install the server into Claude Desktop with the MCP CLI. section 3.2.1
  • Open File > Settings > Developer to verify that the server is running. section 3.2.1
  • Start a conversation, request the tool, and approve its execution when Claude prompts for permission. section 3.2.1

A minimal research-tools connection

A FastMCP server named “Research Tools” registers get_research_sources(), whose signature, docstring, and return annotation become the tool schema Claude uses. Running mcp install 01_claude_mcp_server.py adds the server to Claude’s configuration. After restarting or opening Claude Desktop, verify the server in Developer settings and enter get_research_sources in a new conversation. Claude requests permission before executing it. section 3.2.1

How AI Agents in Action, Second Edition covers it

How deep it goes
Explains configuration through tool execution and troubleshooting, including signatures, docstrings, return types, and permission prompts.
Examples
Uses a Python research-tools server, MCP CLI installation, Desktop settings, and an execution-permission prompt.
What it assumes
Assumes a Python file and MCP CLI setup; flags uv, Python terminal configuration, and correct file paths as prerequisites.
What you can do after
Operate/configure Claude Desktop with an MCP server, inspect availability, approve tool execution, and diagnose common setup errors.
How it connects
Ties MCP servers to tools, prompts, resources, LLM tool registration, and client-specific configuration.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of AI Agents in Action, Second Edition. The section links point to the parts of the book it draws on.