The Complete Practical Guide to Instructing LLMs and AI Agents

Read before you buy: the PDF here is a free sample — the first 20 pages of the 452-page book (payments, sales and reviews on this demo site are sample data). Get the full e-book at thenexbooks on Gumroad.

Get reliable results from AI — every time.

The same model that gives one person a brilliant answer gives another a vague, wrong, or unsafe one. Very often the difference is not the model. It is the prompt.

This is the complete, practical guide to prompt engineering: how models read your words, how to write clear and well-structured prompts, and how to build dependable production systems — RAG, tool use, agents, evaluation, security, and cost control. Vendor-neutral: works with Claude, GPT, Gemini, and open-weight models.

✔ 452 pages · 28 chapters + 4 appendices

✔ Hundreds of before → after prompt rewrites that show exactly what changes and why

✔ Reasoning, structured JSON output, long-context grounding, chaining, self-critique

✔ Patterns for summarization, extraction, classification, writing, code, SQL, and translation

✔ RAG, tool/function calling, agents, and context engineering (including MCP)

✔ Evaluation with test sets and LLM-as-judge, prompt-injection defense, cost and latency optimization

✔ 2 complete case studies + a pattern catalog and copy-ready template library

✔ FREE PREVIEW — 20 sample pages: 14 in the image gallery above (including the full table of contents) + 6 more at the bottom of this description

— TABLE OF CONTENTS —

PART I · FOUNDATIONS

Ch 1 · What Prompt Engineering Is (and Isn't)

Ch 2 · How LLMs Work: Tokens, Context Windows, and Sampling

Ch 3 · Anatomy of a Prompt: System, User, and Assistant Messages

Ch 4 · Clarity and Specificity: Writing Instructions Models Can Follow

Ch 5 · Learning from Examples: Zero-Shot, One-Shot, and Few-Shot

Ch 6 · Structuring Prompts: Delimiters, XML Tags, and Templates

PART II · CORE TECHNIQUES

Ch 7 · Reasoning: Chain-of-Thought and Reasoning Models

Ch 8 · Roles, Personas, Tone, and Audience

Ch 9 · Structured Output: JSON, Schemas, and Validation

Ch 10 · Working with Long Context: Documents, Grounding, and Citations

Ch 11 · Decomposition and Prompt Chaining

Ch 12 · Self-Consistency, Self-Critique, and Reflection

PART III · APPLICATIONS

Ch 13 · Summarization and Information Extraction

Ch 14 · Classification, Routing, and Moderation

Ch 15 · Writing, Editing, and Style Control

Ch 16 · Code Generation, Review, and Debugging

Ch 17 · Data Analysis and Text-to-SQL

Ch 18 · Multilingual Prompting and Translation

PART IV · RETRIEVAL, TOOLS, AND AGENTS

Ch 19 · Retrieval-Augmented Generation (RAG)

Ch 20 · Tool Use and Function Calling

Ch 21 · Building Agents: Loops, Planning, and Memory

Ch 22 · Context Engineering for Agents (including MCP)

PART V · PRODUCTION

Ch 23 · Evaluating Prompts: Test Sets, Metrics, and LLM-as-Judge

Ch 24 · Prompt Security: Injection, Jailbreaks, and Guardrails

Ch 25 · Cost, Latency, and Reliability

Ch 26 · PromptOps: Versioning, Templates, Monitoring, and Iteration

PART VI · CASE STUDIES

Ch 27 · The Orbit Store Support Assistant

Ch 28 · An Agentic Catalog and Operations Pipeline

APPENDICES

A · Prompt Pattern Catalog · B · Template Library · C · Production Checklist · D · Glossary

— WHO IT IS FOR —

Developers, product builders, analysts, writers, and team leads who use large language models through chat or APIs and want consistently better, safer results.

— FORMAT —

PDF (A4), 452 pages, English, with a clickable table of contents and PDF bookmarks. 1st Edition, 2026.

— MORE SAMPLE PAGES (pages 15–20 of the free preview) —

Book details