Automates linear multi-agent pipelines where outputs from one agent flow seamlessly into the next stage.
This skill enables the creation of structured, sequential agent workflows using AG2 (formerly AutoGen). It allows developers to define a series of distinct stages—such as research, drafting, and formatting—where each agent builds upon the previous one's output using a predictable execution path. By utilizing the asynchronous a_sequential_run pattern, it provides a robust framework for complex tasks that require clear handoffs, ensuring that multi-step processes like content pipelines or ETL workflows are handled with precision and context awareness.
主な機能
01Support for Claude-native LLMConfig for optimized performance
02Scalable architecture for multi-step transformation pipelines
03Automated context passing through stage-to-stage summary methods
042 GitHub stars
05Configurable agent handoffs with defined input/output expectations
06Asynchronous workflow execution using the a_sequential_run pattern
