Appendix D. Reference documents

All pages below were checked on 2026-10-20. Each entry names the chapters using it. Documentation changes faster than this book; use each page's latest revision while studying.

Anthropic

• Building effective agents — Chapters 1 and 2
• Writing effective tools for agents — Chapters 2 and 4
• How we built our multi-agent research system — Chapters 1, 3, 4 and 5
• Effective context engineering for AI agents — Chapter 3
• Effective harnesses for long-running agents — Chapter 3
• Demystifying evals for AI agents — Chapter 4
• Claude API: Prompt caching — Chapter 3
• Claude API: Context editing — Chapter 3
• Claude API: Compaction — Chapter 3
• Claude API: Memory tool — Chapter 3
• Claude API: Errors — Chapter 5
• Claude API: Rate limits — Chapter 5
• Claude Code: Hooks reference — Chapter 5
• Claude Code: Configure permissions — Chapter 5
• Claude Code: Sandboxing — Chapter 5

Model Context Protocol

• Specification 2026-07-28 — Chapter 2
• Base protocol — Chapters 2 and 4
• Key changes — Chapter 2
• Deprecated features — Chapter 2
• Tools — Chapter 2
• Schema reference (ToolAnnotations) — Chapter 2
• Multi round-trip requests — Chapter 2
• Cancellation and timeouts — Chapters 2 and 5
• Elicitation — Chapter 2
• Security best practices — Chapter 2
• Client best practices — Chapter 2
• Tasks extension — Chapter 2

OpenAI

• A practical guide to building agents — Chapters 1 and 5
• Agents SDK: Agent orchestration — Chapter 1
• Agents SDK: Running agents — Chapters 3 and 5
• Agents SDK: Sessions — Chapter 3
• Agents SDK: Guardrails — Chapter 5
• Agents SDK: Human-in-the-loop — Chapters 3 and 5
• Agents SDK: Tracing — Chapter 4

Google

• ADK: Workflow agents — Chapter 1
• ADK: Why evaluate agents — Chapter 4
• ADK: Evaluation criteria — Chapter 4
• ADK: Callbacks — Chapter 5
• ADK: Plugins — Chapter 5
• ADK: State — Chapter 3
• ADK: Memory — Chapter 3

OWASP, NIST, AWS

• OWASP Top 10 for LLM Applications 2025 — Chapter 5
• OWASP LLM01:2025 Prompt Injection — Chapter 5
• OWASP LLM06:2025 Excessive Agency — Chapter 5
• OWASP LLM10:2025 Unbounded Consumption — Chapter 5
• NIST AI Risk Management Framework — Chapter 5
• Amazon Builders' Library: Timeouts, retries and backoff with jitter — Chapter 5