Chapter 11 Further Reading: Prompt Engineering Patterns

Books

"The Pragmatic Programmer" (20th Anniversary Edition) David Thomas and Andrew Hunt (Addison-Wesley, 2019)

While not about AI, this foundational software engineering book contains the clearest articulation of the "don't repeat yourself" principle and the value of building reusable patterns vs. solving the same problem repeatedly. Chapter 7 on "coding practices" applies directly to the practice of building prompt libraries: the same reasoning that motivates reusable code applies to reusable prompts. Recommended for any reader who wants the deep intellectual foundation for pattern-based thinking.

"Design Patterns: Elements of Reusable Object-Oriented Software" Gamma, Helm, Johnson, and Vlissides (Addison-Wesley, 1994)

The original "Gang of Four" design patterns book. Again, not about AI prompting — but the structure it uses (pattern name, intent, applicability, structure, participants, consequences) is the exact model for documenting prompt patterns well. The way this book catalogs 23 software patterns is a template for how to catalog prompt patterns. Reading the introduction alone, and understanding what makes a good software pattern, illuminates what makes a good prompt pattern.

"Atomic Habits" James Clear (Avery, 2018)

The most useful book on building systematic behavior — including the habit of building and using prompt templates rather than defaulting to ad hoc approaches. Clear's concept of "environment design" (making the right behavior the easy behavior) applies directly: storing patterns where you actually work makes pattern use the path of least resistance.

"The Checklist Manifesto" Atul Gawande (Metropolitan Books, 2009)

Gawande's study of how checklists improve performance in high-stakes professional contexts — surgery, aviation, construction — maps directly onto prompt patterns. Both are mechanisms for making expert knowledge systematic, accessible, and consistently applied. The book is a compelling case for why even experts benefit from structured templates. Particularly relevant to Elena's consulting pattern library.

Online Resources

"Prompt Engineering Guide" — Patterns and Templates Section https://www.promptingguide.ai/techniques

The DAIR.AI Prompt Engineering Guide's section on prompting techniques includes practical examples of template-based prompting. The guide is updated regularly, making it a reliable reference as best practices evolve. Start with the "few-shot prompting" and "zero-shot prompting" sections, then move to the advanced techniques for pattern composition.

"Awesome ChatGPT Prompts" (GitHub) https://github.com/f/awesome-chatgpt-prompts

A community-curated list of high-quality prompts across hundreds of use cases, organized by role and task type. Useful as a source of inspiration and comparison: after building your own patterns, review how others have approached the same task types. Many prompts in this repository are implicitly template-structured; the practice of converting specific prompts in this list into proper [BRACKET VARIABLE] patterns is a useful exercise.

Notion AI Templates Gallery https://www.notion.so/templates/categories/ai

Notion's template library includes AI prompt templates for many professional use cases. Useful for: (1) seeing how others structure recurring task templates, (2) finding starting-point templates to adapt for your own library. The templates here tend to be simpler than the patterns in this chapter but are immediately usable.

Anthropic Claude Prompt Library https://docs.anthropic.com/en/prompt-library/library

Anthropic's official collection of high-quality prompts for professional use cases. Each entry is effectively a documented pattern: it has a title, use case, and full prompt. Studying how these are structured is directly applicable to building your own pattern library. Many of the prompts can be adapted into [BRACKET VARIABLE] templates with minimal effort.

Research and Theory

"Template-Based Question Answering Using Natural Language Processing" (Various authors, proceedings from NLP research conferences)

The academic literature on template-based NLP is voluminous and pre-dates the current LLM era. For practitioners rather than researchers, the key insight is that structured templates have consistently outperformed unstructured approaches for reliability and consistency in language tasks — a finding that directly motivates the chapter's approach to prompt patterns.

"Cognitive Offloading" Research Risko, E. F., & Gilbert, S. J. (2016). "Cognitive Offloading." Trends in Cognitive Sciences.

The psychology research on cognitive offloading — using external tools and structures to reduce the burden on working memory — provides the theoretical underpinning for why prompt libraries work cognitively, not just practically. Templates offload the cognitive work of prompt construction so working memory can be applied to the actual task. Relevant for anyone who wants the research basis for the "reducing cognitive load" benefit described in the chapter.

"Knowledge Management and Organizational Performance" Various sources in the knowledge management literature

The idea that explicit templates capture and transfer tacit expertise has a long history in knowledge management research (Nonaka's work on knowledge conversion is a starting point). Prompt pattern libraries are a new instantiation of this idea: they convert the tacit knowledge of "how to prompt for this task type" into explicit, transferable templates. For practitioners in knowledge-intensive professions (consulting, law, research), this framing is particularly relevant.

Tools for Building and Managing Pattern Libraries

Notion https://www.notion.so

The most popular choice for knowledge workers building prompt libraries. Supports template databases, tagging, linking, and version history. Useful template: a database with fields for Pattern Name, Use Case, Template Text, Last Updated, and Notes. The template function within Notion allows you to create new pattern entries from a pre-built structure.

Obsidian https://obsidian.md

A local-first note-taking tool with excellent search and linking. Preferred by users who want their pattern library stored locally (not in the cloud), or who want to link patterns to related notes and examples. The "daily note" functionality can be adapted for post-task pattern refinement notes.

GitHub / GitLab (for technical teams) For engineering teams, storing prompt patterns in a code repository enables: version control with history of changes, team collaboration via pull requests, and integration with existing engineering documentation workflows. A repository of markdown files with pattern documentation, reviewed via PR before being added to the official library, is a natural fit for teams already working this way.

Custom GPTs (OpenAI) https://chat.openai.com/gpts

OpenAI's Custom GPT functionality allows you to create a custom AI assistant with your patterns pre-loaded in the system prompt. This eliminates the need to paste templates each time — you open your custom GPT and it already knows your patterns. Best suited for your 3-5 most frequently used patterns. Not suited for patterns that require large amounts of context (the system prompt has limits) or patterns you update frequently.

A Note on This Rapidly Evolving Space

The tools for building and managing prompt libraries are evolving faster than any other aspect of AI workflow. As of early 2026, new tools specifically designed for prompt management are emerging: dedicated prompt management platforms, IDE extensions for developers, and workflow automation tools that treat prompts as versioned assets.

The principles in this chapter are tool-agnostic and will remain applicable as the landscape evolves. Focus on building the patterns (the intellectual work) rather than optimizing the storage tool. The right tool is the one you will actually use; switch to a better tool when the friction of the current one becomes noticeable, not before.