when-chaining-agent-pipelines-use-stream-chainlisted

# Agent Pipeline Chaining SOP ## Overview This skill implements agent pipeline chaining where outputs from one agent become inputs to the next, supporting both sequential and parallel execution patterns with streaming data flows. ## Agents & Responsibilities ### task-orchestrator **Role:** Pipeline coordination and orchestration **Responsibilities:** - Design pipeline architecture - Connect agent stages - Monitor data flow - Handle pipeline errors ### memory-coordinator **Role:** Data flow and state management **Responsibilities:** - Store intermediate results - Coordinate data passing - Manage pipeline state - Ensure data consistency ## Phase 1: Design Pipeline ### Objective Design pipeline architecture with stages, data flows, and execution strategy. ### Scripts ```bash # Design pipeline architecture npx claude-flow@alpha pipeline design \ --stages "research,analyze,code,test,review" \ --flow sequential \ --output pipeline-design.json # Define data flow npx claude-flow@alpha pipeline dataflow \ --design pipeline-design.json \ --output dataflow-spec.json # Visualize pipeline npx claude-flow@alpha pipeline visualize \ --design pipeline-design.json \ --output pipeline-diagram.png # Store design in memory npx claude-flow@alpha memory store \ --key "pipeline/design" \ --file pipeline-design.json ``` ### Pipeline Patterns **Sequential Pipeline:** ``` Agent1 → Agent2 → Agent3 → Agent4 ``` **Parallel Pipeline:** ``` ┌─ Agent2 ─┐ Agent1 ├─ Age