Everything you need to know about the Skills for Real Engineers repository, from installation to advanced usage.
These are agent skills designed specifically for real engineering work. They're small, composable tools that help you work more effectively with AI coding assistants like Claude Code. Unlike traditional approaches like GSD or BMAD, these skills give you full control while providing structure and best practices.
Installation is simple:
npx skills@latest add mattpocock/skillsThe setup will ask about your issue tracker (GitHub, Linear, or local files), triage labels, and documentation preferences.
The skills work with any AI coding agent and require Node.js (v14+) for the installer. Once installed, the skills themselves are lightweight and work entirely within your agent's environment. No additional dependencies are needed for the skills to function.
User-invoked skills are triggered when you type their commands (like /grill-me). They orchestrate workflows and guide conversations.
Model-invoked skills can be triggered by you or automatically by the agent when the task fits. They contain reusable disciplines and best practices.
User-invoked skills can call model-invoked skills, but not other user-invoked skills.
The skills are language-agnostic. They focus on engineering principles and processes rather than language-specific syntax. The /tdd skill, for example, works with any testing framework, and the /diagnose skill applies debugging techniques universally.
Absolutely! The skills are designed to be adaptable. You can modify the prompts, adjust the workflows, and integrate them with your existing tools. The /setup-matt-pocock-skills lets you configure issue trackers, labels, and documentation paths to match your team's process.
This skill conducts a comprehensive grilling session while building your project's domain model. It challenges your plans against existing terminology, sharpens definitions, and automatically updates your CONTEXT.md and ADRs. It's particularly powerful for reducing verbosity and creating a shared language between you and the AI.
The /tdd skill enforces a red-green-refactor loop. It encourages the AI to write failing tests first, then implement the code to pass those tests. This creates immediate feedback loops, reduces bugs, and results in better-designed code. It builds features one vertical slice at a time for consistent quality.
This skill combats codebase entropy by identifying opportunities to deepen modules. It scans your code for areas where you can add more functionality behind simpler interfaces, based on your domain language and architectural decisions. It presents findings as a visual HTML report and guides you through improvements.
First, check your installation by running /setup-matt-pocock-skills again. If issues persist, verify your agent is compatible with the skills. For specific skill problems, check the skill documentation in the GitHub repo. You can also open an issue on GitHub with details about your environment and the problem.
Bugs and improvement suggestions should be reported as issues on the GitHub repository. Please include:
The best place for community support is the GitHub Discussions tab. You can also join the newsletter for updates and tips. Many users share their experiences and solutions in the GitHub issues and pull requests.
The repository is licensed under the MIT License, which is very permissive. You can freely use, modify, and distribute the skills in both personal and commercial projects, as long as you include the original license and copyright notice.
The skills are actively maintained based on real-world usage and feedback. New skills are added regularly, and existing ones are improved as we learn more about effective AI-assisted development. The repository has received 147,154 stars from developers worldwide, indicating strong community interest and validation.
The skills themselves are completely free and open-source. Any costs would come from the AI coding agent you choose to use them with (like Claude Code). The skills are designed to maximize the value you get from these tools by improving efficiency and code quality.