pipeline

You can orchestrate multi-stage AI workflows by configuring a pipeline with built-in stages like ralplan, team-exec, and ralph-verify. The pipeline persists durable state under .omq/state, enabling automatic resume and retry across runs after interruptions.

To set up an end-to-end AI task pipeline in Qwen Code, create an autopilot pipeline configuration that sequences planning, execution, and verification stages. You can configure iteration limits and worker counts for built-in stages to customize the workflow.

Yes, you can build custom stages by implementing the uniform PipelineStage interface. This interface ensures consistent stage execution and easy extension, allowing you to integrate custom logic into your existing autopilot, team, or ralph workflows.

The best way to automate planning and execution sequences is using a pipeline orchestration tool that coordinates stages and persists state. This approach integrates seamlessly with autopilot and team modes to manage end-to-end workflows efficiently.

Yes, pipeline orchestration supports state management for auditing AI runs by persisting durable state under the .omq/state directory. This mechanism tracks workflow progress across runs, supporting both auditing and recovery after failures.