arlandaren/proagents

Proagents is a library of 794 pre-written system prompts, agent personas, and task workflows for AI coding tools. The idea is that instead of writing your own instructions each time you open a project in an AI editor like Cursor, Claude Code, or Windsurf, you can install a ready-made expert profile that tells the AI how to behave as a particular kind of specialist. A simple command-line tool lets you browse, search, and install these prompts into your editor's configuration file. The library is organized by domain. There are 126 engineering personas covering roles like senior full-stack developer, mobile app specialist, game engine developer, and embedded firmware engineer. There are 15 design personas for UI designers, UX researchers, and brand specialists. There are 33 operations-focused personas for code reviewers in various languages, database reviewers, and technical onboarding guides. The library also includes 521 workflow checklists covering things like test-driven development steps, security review processes, deployment procedures, and API integration guides. No dependencies are required to use the CLI, it runs on Python 3.8 and above. The prompts are described as distilled from six existing open-source agent frameworks and tested across real projects rather than written speculatively. The README compares the default AI assistant output to what the library's prompts are said to produce: generic layout versus curated design decisions, missing edge cases versus strict validation, skipped tests versus test-first workflow, and so on. Whether a specific prompt delivers those improvements depends on the AI model and context you are using it in. Installation is a single clone with no pip install. From there you use a CLI script to list available domains, search by keyword, and install a prompt to your editor's rules file or print it to standard output so you can append it manually. The project is released under the MIT license and available in English, Russian, and Chinese.