In a Swarm / Team Agent Setup, where you have defined Specialists, you define in short and compressed when possible, the Skills, this Sub-Agent is gonna Pickup. In this Example, the Skills are named identically as the executing specialist / sub-agent.
As written on the Oveview page for this example project, you may define certain skills more in depth, but usually only to prevent cavecats or point to specific rules for a specific file or folder/component/node or functions. The LLM only needs Informations, which are hard to guess or easy to miss.
The Skills above are primary for an entire refactoring with code optimization both for security and performance. You noticed, each skill is around 80 - 180 Lines only. And that’s the ideally amount of information to be in a single skill file.
But if you want to have a bit more Security on Compiling, and creating an App Version, you may want to use PVYlima Desktop VM for Comppiling, and maybe also the PVYlima VM “Understand Anything” which you either let configure through the LLM or manually by your self to scan the right Repository.
So Further Skills could be:
Which covers how the Sub-Agent shall compile NodeJS, TypeScript and Electron, using an isolated VM over PVYlima.
Beside of compiled Project Output Folder, where he can find the Apple Developer Certificate, the KEY and Apple SKU and Profile Information for this Project, and how to compile it for other OS Platforms.