Research pipelines
Multi-agent orchestration, state recovery, and context-efficient hand-offs.
15 questions in the full mock. Strong-domain archetype: Orchestration Lead. This is an unofficial study aid by CyberSkill, not affiliated with or endorsed by Anthropic.
What it covers
Research-pipeline questions put you in charge of a multi-agent system that gathers and synthesizes information. You decide how to split work across agents, how to hand off context between them without wasting tokens, and how to recover state when a sub-agent fails partway through. Expect scenarios about orchestration patterns, retries, and keeping a long-running job coherent.
Why it matters for an agent architect
Multi-agent research is where small design mistakes compound fastest. An orchestration lead who plans for failure, keeps hand-offs lean, and can resume after a crash ships systems that finish their work instead of stalling or burning context. This domain checks that judgment directly.
Drill this domain
Work the 15 questions in research pipelines on their own until the reasoning clicks, then sit the full timed mock to see how it holds up under the clock.