CoPrompt is a collaborative prompt engineering tool designed to assist programmers in natural language programming by facilitating seamless collaboration on prompt creation and refinement. It offers four key mechanisms: sharing, referring, linking, and requesting. These features enable programmers to share information effortlessly, modify prompts by referring to others' work, link prompts for automatic updates, and request assistance without interrupting collaborators' workflow. By providing multi-level hierarchical views and contextual prompt information, CoPrompt helps reduce communication barriers and repetitive updates, making collaborative prompt engineering more efficient.
In a user study with 12 experienced programmers, CoPrompt improved both real-time and asynchronous collaboration by helping users easily understand, build upon, and modify collaborators' prompts. It addressed common challenges like maintaining context, tracking revisions, and managing dependencies, making it an effective tool for streamlining collaborative NL programming workflows.
- Core Features
- Seamless collaboration through sharing, referring, linking, and requesting prompts
- Multi-level hierarchical views for prompt comprehension and modification
- Automated updates of linked prompts, reducing repetitive changes
- Facilitates both real-time and asynchronous collaboration
- SkillsPython, PyTorch, scikit-learn, Machine Learning, Evasion Attacks, D3.js, JavaScript, HTML, CSS
- AuthorsLi Feng, Ryan Yen, Yuzhe You, Mingming Fan, Jian Zhao, Zhicong Lu
- Paper LinkCoPrompt: Supporting Prompt Sharing and Referring in Collaborative Natural Language Programming
- KeywordsHCI, Collaborative Programming, Natural Language Programming, Prompt Engineering, Sharing