In this section: Chapters

Making Your System Legible to AI Agents

The technical model for agent-navigable architecture

Chapter 6 is the technical model for a system an agent can navigate. It covers what agents actually need. It covers contract completeness: the minimum viable contract, the extended contract, and anti-patterns. It covers MCP as the runtime bridge, the agent onboarding test, and governing the graph under agent-assisted evolution.

The agent onboarding test checks whether a capability's contract is enough, on its own, for an agent to navigate the capability safely. It evaluates the contract, not the agent. The book's write-up of that test, including scoring bands and a legibility index, isAppendix C.

Forthcoming from Apress. ISBN 9798868834516.