stream-chain

Multi-agent orchestration chains sequential AI steps by streaming the complete output from one agent directly to the next, enabling coordinated workflows without manual handoffs. stream-chain automates this propagation across custom or predefined pipelines, maintaining data consistency throughout the sequence.

Yes. stream-chain coordinates multi-agent pipelines for data transformation by chaining agents in sequence, where each agent processes the previous step's full output. This works for both custom chains you define and battle-tested predefined pipelines.

Streaming JSON chaining passes structured data seamlessly across agent boundaries. stream-chain manages JSON serialization and propagation between sequential agents, eliminating format translation friction and ensuring each agent receives complete context from prior steps.

Chain multiple agents sequentially: research agent outputs topic analysis, which streams to a drafting agent for article creation, then to an optimization agent for refinement. stream-chain automates this multi-step coordination so each agent receives the previous agent's complete output.

Both. stream-chain supports custom chains where you define your own agent sequences and predefined pipelines for common tasks. Custom chains give full control over the workflow, while predefined pipelines accelerate development with battle-tested patterns.

stream-chain requires claude-flow as its primary dependency. This framework provides the foundation for orchestrating agent communication and streaming data flow across pipeline steps.