← All domains

Domain 1 · 27%

Agentic Architecture & Orchestration

Design multi-step Claude agents: when to use loops vs. orchestrator-worker, how to compose subagents, and how to evaluate agent outputs.

Agent vs. Workflow
When to give Claude autonomy vs. when to pin a fixed sequence.
6 min
Orchestrator-Worker Pattern
One Claude routes subtasks to specialized subagents.
8 min
Evaluator–Optimizer Loop
Use Claude to grade Claude's own output and iterate.
7 min
Parallel Fan-Out
Map-reduce over Claude calls when the work shards cleanly.
6 min
Agent State & Memory
Persisting decisions and intermediate results across turns.
9 min
Human-in-the-Loop
Where to surface confirmation prompts without breaking flow.
5 min
Evaluating Agent Quality
Golden tasks, success rates, and step-level traces.
7 min