AI Agent Engineering in Practice: Prompt / Context / Harness Engineering

Backmatter

Backmatter

Reading Guide

This reading guide is for the point after a reader has finished the chapter prose and repo artifacts and wants durable sources that deepen understanding. It favors reuse value over exhaustiveness.

How to Use This Guide

  • Return first to the manuscript and repo artifacts in this book
  • Then read the official docs for the tool or platform you actually use
  • Use books and handbooks to strengthen organizational judgment, not to replace local artifacts
  • The goal is not academic completeness. The goal is durable next-step reading

Source Hierarchy

In the 2026 edition, use sources in the following order.

  1. The manuscript and sample-repo: the canonical source for recurring cases and artifact responsibilities
  2. Official docs and protocol specifications: the primary source for runtime behavior, permissions, pricing, and protocol detail
  3. Books and handbooks: secondary sources that strengthen review, operations, and reliability judgment

If the prose and the live tool differ, keep the design principle from the manuscript and re-read the execution conditions in the primary source.

Keep the naming rule explicit. This guide normally points to categories such as the agent runtime in use, the VCS / CI system in use, and internal policy. Name a specific vendor or product document only when you must confirm API, CLI, permission, pricing, or protocol behavior, and track the confirmation date plus model/runtime profile with the evidence.

Prompts and Requirements Shaping

Source What It Strengthens Related Chapters
Official prompting guide for the model in use Prompt Contracts, constraint design, and tool-use assumptions CH02, CH04
Official eval docs for the model in use Case design, rubrics, and version comparison CH04
Michael Nygard, Architecture Decision Records Writing short and durable design decisions CH03
Internal product-spec and acceptance-criteria templates The threshold for turning exploratory dialogue into implementation-ready artifacts CH03

Context and Repo Design

Source What It Strengthens Related Chapters
Official docs for the agent runtime in use Instruction layering, workspace access, and session handling CH05, CH06, CH07, CH08
Official docs for the VCS, CI, and package manager in use Issues, pull requests, review flow, branch protection, and artifact traceability CH03, CH06, CH10
Internal repo maps, architecture docs, and coding standards Canonical repo context and clearer ownership CH06
Internal handoff, incident, and change-log rules Session memory, restart packet (Resume Packet), Progress Note, handoff contracts, and approval boundary CH07, CH11, CH12

Verification, Reliability, and Operations

Source What It Strengthens Related Chapters
Official docs for the test framework, CI system, and coverage tool in use Verify boundaries, evidence, and execution-condition differences CH09, CH10, CH11, CH12
Betsy Beyer et al., Site Reliability Engineering Reliability, operational responsibility, and service thinking CH09, CH10, CH12
Betsy Beyer et al., The Site Reliability Workbook Checklists, runbooks, and implementation-focused operations design CH09, CH10, CH11, CH12
Nicole Forsgren, Jez Humble, Gene Kim, Accelerate Metrics, throughput, and operational improvement CH10, CH12
Internal approval and permission policy Human approval gates, authority boundaries, and auditability CH09, CH12
Internal PR templates, review checklists, and merge policy Lead / Operator / Reviewer responsibilities, Goal, Changed Files, Scope and Non-goals, Verification, Evidence / Approval, and review-budget operations CH10, CH12
  • If a prompt question is unclear, return first to official docs and local eval artifacts
  • If a context question is unclear, return first to repo artifacts and ownership documents
  • If a harness question is unclear, return first to verify, evidence, approval boundary, and review policy
  • Books and handbooks should reinforce judgment, not replace local artifacts