Prompt Engineering Resources
Academic papers, platform guides, and practitioner references organized by technique. Each pattern page links to its most relevant papers; this page collects them all in one place.
Platform Documentation
Section titled “Platform Documentation”These official guides are maintained by the AI platform teams and reflect current best practices.
| Platform | Guide | Notes |
|---|---|---|
| Anthropic (Claude) | Prompt Engineering Guide | Comprehensive guide covering all major techniques |
| OpenAI (ChatGPT) | Prompt Engineering Best Practices | Strategies and tactics for better results |
| OpenAI (GPT-5.4) | Prompt Guidance for GPT-5.4 | Model-specific guidance for long-running tasks, tool use, and reliable execution |
| Google Cloud | Prompt Engineering Overview and Guide | Overview of prompt engineering concepts and techniques |
Courses and Learning Resources
Section titled “Courses and Learning Resources”| Resource | Provider | Notes |
|---|---|---|
| Prompt Engineering Courses | edX | University-backed courses on prompt engineering |
| Prompt Engineering Courses | Coursera | Courses from industry and academic partners |
| Prompting Guide | DAIR.AI | Open-source guide covering techniques, applications, and research |
Academic Papers by Technique
Section titled “Academic Papers by Technique”Zero-Shot Prompting
Section titled “Zero-Shot Prompting”- Brown et al. 2020 — Language Models are Few-Shot Learners — arxiv.org/abs/2005.14165 — The GPT-3 paper that established the zero-shot paradigm, demonstrating that large language models can perform tasks from instructions alone
- Kojima et al. 2022 — Large Language Models are Zero-Shot Reasoners — arxiv.org/abs/2205.11916 — Showed that adding “Let’s think step by step” enables zero-shot chain-of-thought reasoning
Few-Shot Learning
Section titled “Few-Shot Learning”- Brown et al. 2020 — Language Models are Few-Shot Learners — arxiv.org/abs/2005.14165 — Seminal paper demonstrating few-shot learning via text demonstrations
- Dong et al. 2022 — A Survey on In-context Learning — arxiv.org/abs/2301.00234 — Comprehensive survey on why few-shot demonstrations work and how models learn in context
Chain-of-Thought
Section titled “Chain-of-Thought”- Wei et al. 2022 — Chain-of-Thought Prompting Elicits Reasoning in Large Language Models — arxiv.org/abs/2201.11903 — The seminal chain-of-thought paper showing step-by-step reasoning improves performance on math, logic, and commonsense tasks
- Kojima et al. 2022 — Large Language Models are Zero-Shot Reasoners — arxiv.org/abs/2205.11916 — Zero-shot CoT: “Let’s think step by step”
- Yao et al. 2023 — Tree of Thoughts: Deliberate Problem Solving with Large Language Models — arxiv.org/abs/2305.10601 — Generalizes chain-of-thought into tree-structured reasoning with backtracking
Direct Instruction
Section titled “Direct Instruction”- Wei et al. 2021 — Finetuned Language Models Are Zero-Shot Learners (FLAN) — arxiv.org/abs/2109.01652 — Foundation paper on instruction tuning
- Ouyang et al. 2022 — Training Language Models to Follow Instructions with Human Feedback — arxiv.org/abs/2203.02155 — InstructGPT paper establishing RLHF for instruction following
- Zhang et al. 2023 — Instruction Tuning for Large Language Models: A Survey — arxiv.org/abs/2308.10792 — Survey covering instruction tuning methods and their impact
Contextual Prompting
Section titled “Contextual Prompting”- Dong et al. 2022 — A Survey on In-context Learning — arxiv.org/abs/2301.00234 — Defines the in-context learning paradigm
- Lewis et al. 2020 — Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks — arxiv.org/abs/2005.11401 — RAG framework for augmenting prompts with retrieved context
Role Prompting
Section titled “Role Prompting”- Kong et al. 2023 — Better Zero-Shot Reasoning with Role-Play Prompting — arxiv.org/abs/2308.07702 — Role-play improved ChatGPT accuracy on AQuA from 53.5% to 63.8%
- Zheng et al. 2023 — When “A Helpful Assistant” Is Not Really Helpful — arxiv.org/abs/2311.10054 — Important counterpoint: persona prompts don’t reliably improve performance across all tasks
Output Formatting
Section titled “Output Formatting”- Tam et al. 2024 — Let Me Speak Freely? A Study on the Impact of Format Restrictions on Performance — arxiv.org/abs/2408.02442 — Shows format restrictions can degrade reasoning performance
- Liu et al. 2024 — “We Need Structured Output”: Towards User-centered Constraints on LLM Output — arxiv.org/abs/2404.07362 — User-centered perspective on structured output constraints
Multi-Turn Conversation
Section titled “Multi-Turn Conversation”- Yi et al. 2024 — A Survey on Recent Advances in LLM-Based Multi-turn Dialogue Systems — arxiv.org/abs/2402.18013 — Survey covering dialogue management, context tracking, and coherence
- Zheng et al. 2025 — LLMs Get Lost In Multi-Turn Conversation — arxiv.org/abs/2505.06120 — Documents a 39% average performance drop in multi-turn vs. single-turn interactions
Self-Consistency and Reflection
Section titled “Self-Consistency and Reflection”- Wang et al. 2022 — Self-Consistency Improves Chain of Thought Reasoning in Language Models — arxiv.org/abs/2203.11171 — Sampling diverse reasoning paths and selecting the most consistent answer
- Shinn et al. 2023 — Reflexion: Language Agents with Verbal Reinforcement Learning — arxiv.org/abs/2303.11366 — Verbal self-reflection for iterative improvement
- Madaan et al. 2023 — Self-Refine: Iterative Refinement with Self-Feedback — arxiv.org/abs/2303.17651 — Generate-critique-refine loop without external feedback
Emotional Prompting
Section titled “Emotional Prompting”- Li et al. 2023 — EmotionPrompt: Leveraging Psychology for Large Language Models Enhancement via Emotional Stimulus — arxiv.org/abs/2307.11760 — Showed 10%+ improvement on benchmarks using emotional stimuli, though effects vary by model and task
Reframing Prompts
Section titled “Reframing Prompts”- Ma et al. 2023 — Query Rewriting for Retrieval-Augmented Large Language Models — arxiv.org/abs/2305.14283 — Rewrite-Retrieve-Read framework for query reformulation
Style Unbundling
Section titled “Style Unbundling”- Lenny Rachitsky — Five proven prompt engineering techniques — lennysnewsletter.com — Practitioner origin of the style unbundling technique
- Liu et al. 2023 — Learning to Generate Text in Arbitrary Writing Styles — arxiv.org/abs/2312.17242 — Academic research on style decomposition and reproduction
Summarization and Distillation
Section titled “Summarization and Distillation”- Adams et al. 2023 — From Sparse to Dense: GPT-4 Summarization with Chain of Density Prompting — arxiv.org/abs/2309.04269 — Iterative increasing-density summarization technique
- Jin et al. 2024 — A Comprehensive Survey on Process-Oriented Automatic Text Summarization — arxiv.org/abs/2403.02901 — Broad survey of summarization approaches
Real-World Constraints
Section titled “Real-World Constraints”This is primarily a practitioner pattern. The academic literature on constrained generation focuses on technical format constraints rather than business constraints in prompts. See the platform documentation above for practical guidance.
General References
Section titled “General References”These foundational papers cover topics relevant to multiple techniques.
- Schulhoff et al. 2024 — The Prompt Report: A Systematic Survey of Prompting Techniques — arxiv.org/abs/2406.06608 — Comprehensive taxonomy of 58 prompting techniques with a unified terminology
- Zhou et al. 2022 — Large Language Models Are Human-Level Prompt Engineers — arxiv.org/abs/2211.01910 — Automatic prompt optimization (APE)
- White et al. 2023 — A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT — arxiv.org/abs/2302.11382 — Pattern-based approach to prompt engineering, similar to software design patterns