Chains multiple Claude prompts together so each step's output feeds into the next one. You can run custom sequences or use predefined pipelines for common tasks like codebase analysis, refactoring, test generation, and performance optimization. Think of it as a conveyor belt for multi-step workflows where you need structured transformation rather than a single prompt. The pipeline mode gives you ready-made chains (analysis, refactor, test, optimize) while the run mode lets you define your own sequences. Most useful when you have a clear multi-stage process and want each agent to build on what came before, like going from vulnerability scan to categorization to fixes to tests.
npx skills add https://github.com/ruvnet/ruflo --skill stream-chain