Harness Books
Two books on harness engineering that examine the design philosophies behind Claude Code and Codex, covering constraints, query loops, context governance and multi-agent verification.
Harness Books is a pair of books on harness engineering, pursuing the question of what keeps a code-writing model bounded, continuous and accountable once it is placed inside terminals, repositories, permission systems and team workflows. Rather than walking through source code line by line, they focus on how a harness organizes constraints and execution: prompt layering, query loops, permission decisions, context governance, failure recovery, multi-agent verification, local rules and team institutions.
The first book uses Claude Code as its observation target and focuses on runtime structure, examining why a system eventually grows components such as a control plane, a query loop, tool permissions, context governance, recovery paths, multi-agent verification and team rules. The second book compares the harness design philosophies of both Claude Code and Codex.
Both are available to read online or download as PDF, with a Chinese-language edition of the README also provided. The repository is written in Python.