A centralized, versioned collection of reusable prompts for AI-assisted software development.

This repository is designed to act as a single source of truth for prompts used across multiple projects and repositories, with a primary focus on JS/TS environments and Playwright-based test automation, but without hard coupling to any specific tooling.

• Provide reusable, composable prompts (personas, rules, workflows)

• Enable consistent AI behavior across projects

• Keep prompts tool-agnostic and IDE-friendly (plain files)

• Allow easy reuse via Git, raw imports, or packaging later

prompts/ ├── meta/ # Personas, global rules, style guides ├── utilities/ # Small, focused helpers (e.g. comment analysis, structure validation) └── workflows/ # Task-oriented prompts (e.g. test generation, reviews) 

• Meta prompts
  Define who the agent is and how it should behave.

• Workflow prompts
  Describe what the agent should do step-by-step.

• Utility prompts
  Narrow, reusable instructions for specific transformations or checks.

Note: If you fork this repository, update the URLs below to point to your fork.

Add this repository as a subtree so prompts exist as real files in your project:

git subtree add --prefix=prompts --squash https://github.com/MSroczynski3/testing-prompts.git main

To update later:

git subtree pull --prefix=prompts --squash https://github.com/MSroczynski3/testing-prompts.git main

Tip: You can customize the --prefix to place prompts in a different folder, e.g. .ai-artifacts, .github/prompts, or any path that fits your project structure.

You can reference individual prompts via raw URLs:

https://raw.githubusercontent.com/MSroczynski3/testing-prompts/main/prompts/meta/plan-drafter.md 

This works well for quick experiments or external tools.

To undo the import, delete the folder and commit:

git rm -r prompts git commit -m "Remove prompts subtree"

• Prompts should be:

  • explicit

  • concise

  • deterministic where possible

• Avoid project-specific details unless clearly documented.

• Prefer composition over monolithic prompts.

• Prompts should be usable both:

  • standalone

  • and as building blocks combined with others

• Run prompts/utilities/structure-analyzer.md on new or modified prompts to validate structural compliance.

This repository is versioned via Git.

Breaking changes to prompt behavior should:

• be intentional,

• and documented in commit messages or release notes (when introduced).

Formal npm / GitHub Package distribution may be added later.