Pipeline Pattern

The pipeline pattern chains agent steps in a fixed sequence where each step's output becomes the next step's input, creating a predictable, linear workflow for tasks with well-defined stages. Unlike the orchestrator pattern, where the directing agent decides the order dynamically, pipelines enforce a predetermined sequence such as lint, then test, then fix, then verify, making them more deterministic and easier to monitor in production. The key limitation practitioners hit is that pipelines cannot handle dynamic branching: the moment your task requires a conditional decision mid-flow, such as "if tests pass, proceed; if they fail, diagnose the error and retry," the fixed sequence breaks down and you need an orchestrator or state machine instead.