ccw-chain

Multi-step workflow orchestration is automated by routing natural-language requests to the correct chain pipeline, loading required nodes, and executing tasks end-to-end with state propagation. It matches task types and keywords to choose the best workflow path automatically.

Intent routing works by matching user task types like bugfix, hotfix, exploration, and issue workflows to specific chain triggers. The system reads loaded step docs and assembles Skill calls with correct session handling to execute the appropriate pipeline.

Yes, you can execute a Skill pipeline without manual confirmation by using the -y auto-mode confirmation injection. This feature propagates auto-confirm across the chain execution to ensure coherent multi-step outcomes without interruption.

Progressive chain loading is a mechanism that progressively loads nodes and inspects the workflow graph to execute steps in order. It ensures chain_loader-driven orchestration maintains session-aware argument assembly throughout the pipeline.

Variables are passed between parent and child tasks through delegation and variable propagation. The system continues execution across delegated chain depth while passing and receiving variables to maintain state for coherent multi-step outcomes.

Yes, chain-based task routing supports file-based analyze and debug flows. It implements intent-driven task types that route these specific flows through progressive chain loading to execute the required analysis pipeline.