Skip to content

Prompt engineering templates and guidance for AI-assisted scientific software development, based on "Ten Simple Rules for AI-Assisted Coding in Science" by Bridgeford et al. (2025).

License

Notifications You must be signed in to change notification settings

sscivier/prompt-protocols

Prompt Protocols

Structured prompt templates for AI-assisted scientific software development

This repository provides prompt engineering templates and guidance for scientific researchers who write software as part of their research. Based on "Ten Simple Rules for AI-Assisted Coding in Science" by Bridgeford et al. (2025), these templates help you develop better prompts for AI coding assistants, leading to more reliable and scientifically sound code.

Quick Start

  1. Clone the repository to a central location:

    git clone git@github.com:sscivier/prompt-protocols.git ~/path/to/tools/prompt-protocols
  2. Symlink into your project:

    cd ~/path/to/your-project
    mkdir -p ai-dev
    cd ai-dev
    ln -s ~/path/to/tools/prompt-protocols/prompts ./prompts
    ln -s ~/path/to/tools/prompt-protocols/assessment ./assessment
  3. Add to .gitignore (if using git):

    echo "ai-dev/" >> .gitignore
  4. Start using templates - see SETUP.md for detailed instructions.

What's Inside

  • Prompt Templates - Structured templates for domain research, problem framing, solution specification, test design, code review, and more
  • Assessment Tools - Framework for evaluating prompt quality and effectiveness, that can be given to an AI assistant or used as a self-check
  • Reference Guide - Shortened version of "Ten Simple Rules for AI-Assisted Coding in Science"

Example Workflow

Suppose you need to implement a data processing function but aren't sure about domain standards:

  1. Research: Use 01-domain-research.md to understand standard approaches
  2. Frame: Use 02-problem-framing.md to architect your solution
  3. Specify: Use 05-solution-specification.md to define clear requirements
  4. Test: Use 06-test-design.md to specify test cases before implementation
  5. Implement: Use implementation.md to create your final prompt
  6. Assess: Have AI review your prompt against prompt-assessment.md
  7. Revise: Incorporate AI feedback into your prompt
  8. Run: Run prompt with your AI assistant
  9. Review: Carefully review all AI-generated code, use 08-code-review.md, 09-code-refinement.md, and 10-debugging.md as appropriate

See TEMPLATES-INDEX.md for all available templates and detailed usage guidance.

Setup

Full setup instructions: SETUP.md

Key points:

  • Clone once to a central location
  • Symlink into individual projects' ai-dev/ directory
  • Add ai-dev/ to .gitignore to keep project repos clean
  • Updates propagate automatically across all projects

Contributing

We welcome contributions from researchers! See CONTRIBUTING.md for:

  • How to improve templates
  • Adding domain-specific examples
  • Contribution workflow and guidelines

Please read our Code of Conduct before contributing.

Citation

To cite the Bridgeford et al. (2025) paper, use:

@misc{bridgeford2025simplerulesaiassistedcoding,
      title={Ten Simple Rules for AI-Assisted Coding in Science}, 
      author={Eric W. Bridgeford and Iain Campbell and Zijao Chen and Zhicheng Lin and Harrison Ritz and Joachim Vandekerckhove and Russell A. Poldrack},
      year={2025},
      eprint={2510.22254},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2510.22254}, 
}

License

Apache-2.0. See LICENSE for details.

About

Prompt engineering templates and guidance for AI-assisted scientific software development, based on "Ten Simple Rules for AI-Assisted Coding in Science" by Bridgeford et al. (2025).

Topics

Resources

License

Code of conduct

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •