Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Wordflow: Social Prompt Engineering for Large Language Models

Zijie J. Wang, Aishwarya Chakravarthy, David Munechika, Duen Horng Chau

TL;DR

The paper addresses the challenge of prompting large language models for non-expert users, noting that existing tools mainly target developers. It introduces Wordflow, an open-source browser-based platform that enables social prompt engineering through four integrated views: an editor for text and prompt execution, a local prompt library, a community prompt hub, and a settings panel for LLM selection and output handling. Key contributions include prompt templating with input placeholders, regex-based output parsing, and on-device LLM execution via WebGPU, all within a progressive web app framework; it also demonstrates practical usage via two scenarios in technical writing and translation. The work underscores the potential to democratize LLM use, preserve privacy by enabling local inference, and foster collaborative prompt discovery, while outlining future work on workflow integration, community engagement, and responsible AI safeguards.

Abstract

Large language models (LLMs) require well-crafted prompts for effective use. Prompt engineering, the process of designing prompts, is challenging, particularly for non-experts who are less familiar with AI technologies. While researchers have proposed techniques and tools to assist LLM users in prompt design, these works primarily target AI application developers rather than non-experts. To address this research gap, we propose social prompt engineering, a novel paradigm that leverages social computing techniques to facilitate collaborative prompt design. To investigate social prompt engineering, we introduce Wordflow, an open-source and social text editor that enables everyday users to easily create, run, share, and discover LLM prompts. Additionally, by leveraging modern web technologies, Wordflow allows users to run LLMs locally and privately in their browsers. Two usage scenarios highlight how social prompt engineering and our tool can enhance laypeople's interaction with LLMs. Wordflow is publicly accessible at https://poloclub.github.io/wordflow.

Wordflow: Social Prompt Engineering for Large Language Models

TL;DR

The paper addresses the challenge of prompting large language models for non-expert users, noting that existing tools mainly target developers. It introduces Wordflow, an open-source browser-based platform that enables social prompt engineering through four integrated views: an editor for text and prompt execution, a local prompt library, a community prompt hub, and a settings panel for LLM selection and output handling. Key contributions include prompt templating with input placeholders, regex-based output parsing, and on-device LLM execution via WebGPU, all within a progressive web app framework; it also demonstrates practical usage via two scenarios in technical writing and translation. The work underscores the potential to democratize LLM use, preserve privacy by enabling local inference, and foster collaborative prompt discovery, while outlining future work on workflow integration, community engagement, and responsible AI safeguards.

Abstract

Large language models (LLMs) require well-crafted prompts for effective use. Prompt engineering, the process of designing prompts, is challenging, particularly for non-experts who are less familiar with AI technologies. While researchers have proposed techniques and tools to assist LLM users in prompt design, these works primarily target AI application developers rather than non-experts. To address this research gap, we propose social prompt engineering, a novel paradigm that leverages social computing techniques to facilitate collaborative prompt design. To investigate social prompt engineering, we introduce Wordflow, an open-source and social text editor that enables everyday users to easily create, run, share, and discover LLM prompts. Additionally, by leveraging modern web technologies, Wordflow allows users to run LLMs locally and privately in their browsers. Two usage scenarios highlight how social prompt engineering and our tool can enhance laypeople's interaction with LLMs. Wordflow is publicly accessible at https://poloclub.github.io/wordflow.
Paper Structure (12 sections, 6 figures)

This paper contains 12 sections, 6 figures.

Table of Contents

  1. Introduction
  2. Background & Related Work
  3. System Design & Implementation
  4. Editor View
  5. Personal Prompt Library
  6. Community Prompt Hub
  7. Open-source Implementation
  8. Usage Scenarios
  9. Improving Technical Writing
  10. Customizing Translation Styles
  11. Discussion & Future Work
  12. Conclusion

Figures (6)

  • Figure 1: With Wordflow, users can easily manage and customize their prompts. (A) The Personal Prompt Library provides an overview of local prompts, allowing users to search, sort, and customize the quick-action prompt toolbar in the Editor View. (B) The Prompt Editor, activated by clicking a Prompt Card, employs progressive disclosure to help users edit basic prompt information, advanced settings (e.g., output parsing rules and LLM temperature), and sharing configurations.
  • Figure 2: The Prompt Editor allows users to easily configure LLM settings such as temperature and output parsing rules.
  • Figure 3: Wordflow supports remote and local LLMs.
  • Figure 4: The Prompt Viewer shows detailed information about a community prompt. Users can click a button to copy this prompt into their Personal Prompt Library.
  • Figure 5: Google Doc users can directly use Wordflow's add-on to apply prompts to text within their Google Doc documents.
  • ...and 1 more figures