R. Henney | AI Agent Mastery with Claude AI (2026) [PDF, EPUB, MOBI]
Автор: R. Henney
Издательство: Independently published
Жанр: Компьютерная литература
Язык: Английский
Формат: PDF, EPUB, MOBI
Качество: Изначально электронное (ebook)
Иллюстрации: Отсутствуют
Описание:Build Smart Systems That Plan, Reason, and Execute.
There is a fundamental difference between an application that answers questions and one that solves complex problems. Problem-solving requires understanding a broader goal, breaking it down into actionable sub-tasks, managing dependencies, and recovering gracefully when individual steps fail. AI Agent Mastery with Claude AI provides the definitive roadmap for engineering the architecture of deep reasoning.
Written for intermediate to advanced Python developers, this guide demonstrates how to construct systems that formulate structured plans rather than simply generating prose. Every concept is backed by complete, tested code with no pseudocode or placeholder outputs, allowing you to implement production-grade logic immediately.
The structure follows the logical progression of the subject. Part One establishes how language models plan and reason, and how to design prompts that elicit reliable, deep reasoning rather than shallow responses. Part Two covers planning systems: goal decomposition, task trees, dynamic re-planning, and plan validation. Part Three covers execution engines: code agents, web agents, and data pipeline agents. Part Four covers the mastery-level patterns that distinguish senior agent engineers from beginners: reflection loops, knowledge graph reasoning, and the production architectures that scale.
Every code example in this book is complete, tested, and includes its actual output. No pseudocode, no ellipsis, no outputs that were invented rather than measured. The techniques have been chosen because they work in production, not because they work in research papers. Where a technique is theoretically appealing but practically fragile, this book says so explicitly and offers the alternative that actually holds up.
Core Concepts Covered:
- Deep Reasoning Prompts: Design system prompts that force structured decomposition and Socratic questioning.
- Planning Systems: Build task tree generators, resolve dependencies, and manage topological execution ordering.
- Dynamic Re-Planning: Implement self-correction loops, checkpoint recovery, and failure detection mechanisms.
- Execution Engines: Develop code execution sandboxes, REST API orchestrators, and automated data pipelines.
- Mastery-Level Patterns: Deploy the Critic-Revise pattern and integrate Knowledge Graphs for structured analysis.
- Elevate your engineering skills and build autonomous applications capable of genuine, self-directed problem-solving.
Preface
PART ONE: The Architecture of Reasoning
Chapter 1: How Language Models Plan
Chapter 2: Prompt Architectures for Deep Reasoning
Chapter 3: Chain-of-Thought and Extended Thinking
PART TWO: Planning Systems
Chapter 4: Goal Decomposition and Task Trees
Chapter 5: Dynamic Re-Planning and Self-Correction
Chapter 6: Plan Validation and Verification
PART THREE: Execution Engines
Chapter 7: Code Execution Agents
Chapter 8: Web and API Execution Agents
Chapter 9: File and Data Pipeline Agents
PART FOUR: Mastery-Level Patterns
Chapter 10: Reflection and Self-Improvement Loops
Chapter 11: Knowledge Graphs and Structured Reasoning
Chapter 12: Production Mastery — Advanced Patterns
Conclusion
Appendix A: Planning Prompt Templates
Appendix B: Troubleshooting Reasoning Failures
Appendix C: Quick Reference — Patterns and Anti-Patterns
Appendix D: Glossary of Planning and Reasoning Terms
Скриншоты:
Время раздачи: с 10 до 20 (минимум до появления первых 3-5 скачавших) 
