pith. sign in

arxiv: 2607.00043 · v1 · pith:AH23Z5VZnew · submitted 2026-06-29 · 💻 cs.SE

A Taxonomy of Single-Turn Textual Prompt Patterns

Vennila Sooben , Eugene Syriani This is my paper

Pith reviewed 2026-07-02 20:17 UTC · model grok-4.3

classification 💻 cs.SE
keywords prompt patternstaxonomylarge language modelsprompt engineeringsingle-turn interactionsLLM promptssoftware engineering
0
0 comments X

The pith

A taxonomy organizes 30 unique and canonical prompt patterns for single-turn LLM text interactions along two dimensions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes a taxonomy for prompt patterns in single-turn textual LLM interactions by extracting 30 distinct patterns from existing surveys and catalogs through a reproducible method. These patterns are then organized along two dimensions to create a consistent classification. A sympathetic reader would care because varying definitions of prompt patterns across sources create confusion for developers using LLMs in software engineering and everyday tasks. The taxonomy aims to provide a stable reference that minimizes overlap and supports more systematic prompt engineering.

Core claim

Following a reproducible method, the authors identified 30 unique and canonical prompt patterns for single-turn, text-based interactions with large language models, and organized them along two dimensions to form a taxonomy.

What carries the argument

The taxonomy of 30 canonical prompt patterns organized along two dimensions, extracted and validated from prior surveys and catalogs via a reproducible method.

If this is right

  • Developers gain a reference set of standardized patterns instead of relying on ad hoc prompts.
  • The two dimensions supply a framework for relating different prompt techniques to one another.
  • Subsequent literature can align definitions with these 30 patterns to reduce inconsistency.
  • Prompt management tools can adopt the taxonomy for consistent categorization and retrieval.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The taxonomy could serve as a basis for empirical studies comparing effectiveness across the 30 patterns.
  • The reproducible extraction method could be reapplied to multi-turn or multimodal prompt interactions.
  • Standardized patterns might support the design of automated prompt suggestion systems in IDEs.

Load-bearing premise

The patterns extracted from existing surveys and catalogs are sufficiently distinct and canonical to form a stable taxonomy without significant overlap or omission that would require revision of the two dimensions.

What would settle it

A new comprehensive survey that demonstrates substantial overlap among the 30 patterns or identifies key omitted patterns requiring changes to the two organizing dimensions.

Figures

Figures reproduced from arXiv: 2607.00043 by Eugene Syriani, Vennila Sooben.

Figure 1
Figure 1. Figure 1: Process to extract textual single-turn prompt patterns [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Taxonomy categories and their subcategory. Numbers in parenthesis indicate the number of patterns [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
read the original abstract

Large language models (LLMs) are now widely employed in software development and everyday use. Interacting with LLMs requires crafting prompts, which range from simple ad hoc sentences to extensive, detailed, and structured instructions. Knowledge about prompt engineering has been documented in several surveys and catalogs in the literature. However, the term ``prompt pattern'' is defined differently across sources, and existing works have classified prompt patterns in different ways. In this report, we present a taxonomy of prompt patterns for single-turn, text-based interactions. Following a reproducible method, we identified 30 unique and canonical prompt patterns, organized along two dimensions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper claims that, by following a reproducible method applied to existing surveys and catalogs, it has identified 30 unique and canonical single-turn textual prompt patterns for LLMs and organized them along two dimensions.

Significance. A stable, validated taxonomy of prompt patterns could reduce terminological fragmentation in prompt-engineering research and provide a shared reference for practitioners; however, the absence of methodological transparency prevents assessment of whether the claimed 30 patterns and two dimensions are reproducible or stable.

major comments (2)
  1. [Method] Method section: the abstract asserts a 'reproducible method' that yields exactly 30 unique canonical patterns, yet no selection criteria, deduplication rules, source-inclusion protocol, or inter-rater agreement statistics are supplied; without these, independent reproduction of the set of 30 patterns cannot be performed and the stability of the taxonomy cannot be evaluated.
  2. [Taxonomy] Taxonomy section: the two organizing dimensions are presented without explicit justification against alternative classifications in the cited surveys, nor is any overlap or coverage analysis (e.g., pairwise similarity or omission count) reported; this leaves open whether the dimensions are load-bearing or merely post-hoc.
minor comments (2)
  1. [Abstract] The abstract states that patterns are 'unique and canonical' but the body does not define 'canonical' operationally or contrast it with the source definitions that the introduction acknowledges differ across works.
  2. Table or figure presenting the 30 patterns should include source provenance and any merging decisions so readers can trace each pattern back to the input literature.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. The comments correctly identify gaps in methodological transparency and justification that limit the ability to assess reproducibility. We will revise the manuscript to address both points.

read point-by-point responses

  1. Referee: [Method] Method section: the abstract asserts a 'reproducible method' that yields exactly 30 unique canonical patterns, yet no selection criteria, deduplication rules, source-inclusion protocol, or inter-rater agreement statistics are supplied; without these, independent reproduction of the set of 30 patterns cannot be performed and the stability of the taxonomy cannot be evaluated.

    Authors: We agree that the Method section lacks the necessary detail. The abstract's reference to a 'reproducible method' reflects our internal consolidation process across the cited surveys, but this process was not documented with explicit criteria. In the revision we will expand the Method section with: (1) source-inclusion protocol, (2) deduplication rules that reduced the initial set to 30 canonical patterns, (3) selection criteria for uniqueness, and (4) any consistency checks performed. This will enable independent reproduction and stability evaluation. revision: yes

  2. Referee: [Taxonomy] Taxonomy section: the two organizing dimensions are presented without explicit justification against alternative classifications in the cited surveys, nor is any overlap or coverage analysis (e.g., pairwise similarity or omission count) reported; this leaves open whether the dimensions are load-bearing or merely post-hoc.

    Authors: We acknowledge the absence of explicit justification and coverage analysis. The two dimensions were chosen because they capture recurring distinctions in the surveyed literature, but this rationale was not articulated. In the revision we will add a subsection that (a) compares our dimensions against alternative classifications appearing in the cited works and (b) reports a mapping of all 30 patterns onto the dimensions, including overlap counts and any patterns that fall outside the chosen structure. This will demonstrate whether the dimensions are substantive. revision: yes

Circularity Check

0 steps flagged

No circularity in literature synthesis

full rationale

The paper constructs a taxonomy by synthesizing patterns from external surveys and catalogs via a described method. No equations, fitted parameters, self-definitional reductions, or load-bearing self-citations appear; the 30 patterns and two dimensions are outputs organized from independent sources rather than redefined or forced by the paper's own inputs. The work is self-contained as a classification exercise.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The taxonomy depends on the premise that prompt patterns described across heterogeneous sources can be canonicalized into 30 non-overlapping forms without introducing new entities or free parameters.

axioms (1)
  • domain assumption Existing surveys and catalogs contain prompt patterns that admit a stable canonicalization into a small number of unique forms.
    The method begins by identifying patterns from the literature; this assumption is required for the count of 30 to be meaningful.

pith-pipeline@v0.9.1-grok · 5620 in / 1045 out tokens · 20479 ms · 2026-07-02T20:17:44.548584+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

25 extracted references · 12 canonical work pages · 5 internal anchors

  1. [1]

    Angela Fan, Beliz Gokkaya, Mark Harman, Mitya Lyubarskiy, Shubho Sengupta, Shin Yoo, and Jie M. Zhang. Large language models for software engineering: Survey and open problems. InInternational Conference on Software Engineering: Future of Software Engineering, ICSE-FoSE, pages 31–53. IEEE, 2023

    2023

  2. [2]

    REALM: A dataset of real-world LLM use cases

    JingwenCheng,KshitishGhate,WenyueHua,WilliamYangWang, HongShen, andFeiFang. REALM: A dataset of real-world LLM use cases. InFindings of the Association for Computational Linguistics, pages 8331–8341. ACL, 2025

    2025

  3. [3]

    Unleashing the potential of prompt engineering for large language models.Patterns, 6(6):101260, 2025

    Banghao Chen, Zhaofeng Zhang, Nicolas Langrené, and Shengxin Zhu. Unleashing the potential of prompt engineering for large language models.Patterns, 6(6):101260, 2025

    2025

  4. [4]

    A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT

    Jules White, Quchen Fu, Sam Hays, Michael Sandborn, Carlos Olea, Henry Gilbert, Ashraf Elnashar, JesseSpencer-Smith, andDouglasC.Schmidt. Apromptpatterncatalogtoenhancepromptengineering with chatgpt. Technical Report 2302.11382, arXiv, feb 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  5. [5]

    The Prompt Report: A Systematic Survey of Prompt Engineering Techniques

    Sander Schulhoff, Michael Ilie, Nishant Balepur, Konstantine Kahadze, Amanda Liu, Chenglei Si, YinhengLi,AayushGupta,HyoJungHan,SevienSchulhoff,PranavSandeepDulepet,SauravVidyadhara, Dayeon Ki, Sweta Agrawal, Chau Pham, Gerson Kroiz, Feileen Li, Hudson Tao, Ashay Srivastava, Hevander Da Costa, Saloni Gupta, Megan L. Rogers, Inna Goncearenco, Giuseppe Sarl...

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  6. [6]

    A Systematic Survey of Prompt Engineering in Large Language Models: Techniques and Applications

    Pranab Sahoo, Ayush Kumar Singh, Sriparna Saha, Vinija Jain, Samrat Mondal, and Aman Chadha. A systematic survey of prompt engineering in large language models: Techniques and applications. Technical Report 2402.07927, arXiv, mar 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  7. [7]

    A survey of prompt engineering methods in large language models for different nlp tasks

    Shubham Vatsal and Harsh Dubey. A survey of prompt engineering methods in large language models for different nlp tasks. Technical Report 2407.12994, arXiv, jul 2024

    work page arXiv 2024

  8. [8]

    Harrison, and Anton Dereventsov

    Oluwole Fagbohun, Rachel M. Harrison, and Anton Dereventsov. An empirical categorization of prompting techniques for large language models: A practitioner’s guide. Technical Report 2402.14837, arXiv, feb 2024

    work page arXiv 2024

  9. [9]

    From prompts to templates: A systematic prompt template analysis for real-world llmapps

    Yuetian Mao, Junjie He, and Chunyang Chen. From prompts to templates: A systematic prompt template analysis for real-world llmapps. InInternational Conference on the Foundations of Software Engineering, FSE Companion ’25, pages 75–86. ACM, 2025. 23

    2025

  10. [10]

    Language models are few-shot learners

    TomBrown,BenjaminMann,NickRyder,MelanieSubbiah, JaredDKaplan, PrafullaDhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjam...

    2020

  11. [11]

    Addison Wesley Professional, 1994

    Erich Gamma, Richard Helm, Ralph Johnson, and John Vlissides.Design Patterns: Elements of Reusable Object-Oriented Software. Addison Wesley Professional, 1994

    1994

  12. [12]

    The prompt canvas: A literature-based practitioner guide for creating effective prompts in large language models

    Michael Hewing and Vincent Leinhos. The prompt canvas: A literature-based practitioner guide for creating effective prompts in large language models. Technical Report 2412.05127, arXiv, dec 2024

    work page arXiv 2024

  13. [13]

    TELeR: A general taxonomy of LLM prompts for benchmarking complex tasks

    Shubhra Kanti Karmaker Santu and Dongji Feng. TELeR: A general taxonomy of LLM prompts for benchmarking complex tasks. InFindings of the Association for Computational Linguistics, pages 14197–14203. ACL, 2023

    2023

  14. [14]

    The perfect prompt: A prompt engineering cheat sheet.https://medium.com/ the-generator/the-perfect-prompt-prompt-engineering-cheat-sheet-d0b9c62a2bba , apr 2024

    Maximilian Vogel. The perfect prompt: A prompt engineering cheat sheet.https://medium.com/ the-generator/the-perfect-prompt-prompt-engineering-cheat-sheet-d0b9c62a2bba , apr 2024

    2024

  15. [15]

    How I Won Singapore’s GPT-4 Prompt En- gineering Competition

    Sheila Teo. How I Won Singapore’s GPT-4 Prompt En- gineering Competition. https://towardsdatascience.com/ how-i-won-singapores-gpt-4-prompt-engineering-competition-34c195a93d41 , dec 2023

    2023

  16. [16]

    Costar-a: Apromptingframework for enhancing large language model performance on point-of-view questions

    NzubechukwuC.Ohalete, KevinB.Gittner, andLaurenM.Matheny. Costar-a: Apromptingframework for enhancing large language model performance on point-of-view questions. Technical Report 2510.12637, arXiv, oct 2025

    work page arXiv 2025

  17. [17]

    Improving llm’s response with a new prompt framework rcfor

    Tuan Pham and Tuan Nguyen Ngoc. Improving llm’s response with a new prompt framework rcfor. In Applying New Technology in Green Buildings, pages 666–670. IEEE, 2025

    2025

  18. [18]

    Principled instructions are all you need for questioning llama-1/2, gpt-3.5/4

    Sondos Mahmoud Bsharat, Aidar Myrzakhan, and Zhiqiang Shen. Principled instructions are all you need for questioning llama-1/2, gpt-3.5/4. Technical Report 2312.16171, arXiv, jan 2024

    work page arXiv 2024

  19. [19]

    arXiv preprint arXiv:2311.09277 , year=

    Yew Ken Chia, Guizhen Chen, Luu Anh Tuan, Soujanya Poria, and Lidong Bing. Contrastive chain-of-thought prompting. Technical Report 2311.09277, arXiv, nov 2023

    work page arXiv 2023

  20. [20]

    Cumulative Reasoning with Large Language Models

    Yifan Zhang, Jingqin Yang, Yang Yuan, and Andrew Chi-Chih Yao. Cumulative reasoning with large language models. Technical Report 2308.04371, arXiv, may 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  21. [21]

    Code prompting elicits conditional reasoning abilities in Text+Code LLMs

    Haritz Puerto, Martin Tutek, Somak Aditya, Xiaodan Zhu, and Iryna Gurevych. Code prompting elicits conditional reasoning abilities in Text+Code LLMs. InEmpirical Methods in Natural Language Processing, pages 11234–11258. ACL, 2024

    2024

  22. [22]

    Bounding the capabilities of large language models in open text generation with prompt constraints

    Albert Lu, Hongxin Zhang, Yanzhe Zhang, Xuezhi Wang, and Diyi Yang. Bounding the capabilities of large language models in open text generation with prompt constraints. InFindings of the Association for Computational Linguistics, EACL, pages 1982–2008. ACL, 2023. 24

    1982

  23. [23]

    Chatlaw: A multi-agent legal assistant based on a role-aligned mixture-of-experts architecture

    Jiaxi Cui, Munan Ning, Zongjian Li, Hao Li, Yang Ya, Bohua Chen, Bin Ling, Yonghong Tian, and Li Yuan. Chatlaw: A multi-agent legal assistant based on a role-aligned mixture-of-experts architecture. Fundamental Research, 2026

    2026

  24. [24]

    Language Models (Mostly) Know What They Know

    Saurav Kadavath, Tom Conerly, Amanda Askell, Tom Henighan, Dawn Drain, Ethan Perez, Nicholas Schiefer, Zac Hatfield-Dodds, Nova DasSarma, Eli Tran-Johnson, Scott Johnston, Sheer El-Showk, AndyJones, NelsonElhage, TristanHume, AnnaChen, YuntaoBai, SamBowman, StanislavFort, Deep Ganguli, Danny Hernandez, Josh Jacobson, Jackson Kernion, Shauna Kravec, Liane ...

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  25. [25]

    Rcot: Detecting and rectifying factual inconsistency in reasoning by reversing chain-of-thought

    Tianci Xue, Ziqi Wang, Zhenhailong Wang, Chi Han, Pengfei Yu, and Heng Ji. Rcot: Detecting and rectifying factual inconsistency in reasoning by reversing chain-of-thought. Technical Report 2305.11499, arXiv, oct 2023. 25

    work page arXiv 2023