Loop Engineering

A Better Way to Think, Create, and Work with AI - Companion site

← All chapters

Chapter 10 - The Nervous System: MCP and Protocol Interop

Last updated 2026-07-13. Review the protocol-side primary sources behind Chapter 10 and explore the live MCP server catalog for current implementations.

Companion activity

Use the MCP catalog to evaluate one server by asking what it exposes, who authorizes it, what gets logged, and what judgment still lives outside the protocol.

What this chapter argues

MCP - the Model Context Protocol - is one exchange layer through which AI applications can discover and invoke tools or access context. It does not create judgment, governance, security, or learning on its own. The nervous-system analogy is useful but limited: hosts and implementations still decide what is exposed, authorized, logged, reviewed, and retained. A defensible decision record therefore depends on explicit logging and decision context around tool use, not on protocol traffic alone.

Key sources

Chapter 10 cites in the print book: D6, D61, D62, D63. Entries below are the sources this page draws on; some are related literature cited elsewhere in the book rather than in this chapter.

Primary-source links

Chapter contents

Live MCP catalog

The current MCP server catalog - which vendors implement what, with which capabilities, and which client surfaces consume it - lives at /mcp/. Each entry carries a verified date. Refreshed quarterly.