Loop

Agent Patterns

Every core pattern from Anthropic's agent guide, each animated as a live graph.

0 of 8 lessons complete · 84 min total

Start module
  1. The Augmented LLM & Workflows vs AgentsThe building block (LLM + tools + retrieval + memory), the workflow-agent spectrum, and the start-simple rule: only add autonomy when the task demands it.Lesson · 16 minNot completed
  2. Prompt ChainingDecomposing a task into a fixed sequence of LLM calls.Animation · 10 minNot completed
  3. RoutingClassify first, then send to a specialized handler.Animation · 9 minNot completed
  4. Parallelization: Sectioning & VotingSplitting work across simultaneous calls, or asking many times and voting.Animation · 10 minNot completed
  5. Orchestrator-WorkersA lead model decomposes dynamically and delegates.Animation · 10 minNot completed
  6. Evaluator-OptimizerGenerate, critique, refine — in a loop.Animation · 9 minNot completed
  7. The Autonomous Agent LoopGather context, act via tools, verify, repeat — until done or stuck.Animation · 10 minNot completed
  8. Agent Patterns CheckpointMatching problems to patterns and spotting over-engineered autonomy.Checkpoint · 10 minNot completed