Which two tasks are better suited to dynamic adaptive decomposition than to a fixed prompt chain? Select two.

Answer: C and D — Finding out why API latency doubled after last week's deployment / Adding tests to a legacy service whose dependencies are not yet known

Open-ended investigations, where the next step depends on what is found, suit dynamic decomposition that generates subtasks from intermediate findings (c, d). Predictable multi-aspect work with known, repeatable steps, like a standard review checklist (a) or fixed-field extraction (b), suits prompt chaining.