AI Agent Orchestration FAQs

Question 1

What is the difference between an LlmAgent and a WorkflowAgent?

Accepted Answer

LlmAgents use dynamic reasoning to decide which tools to call based on the prompt, whereas WorkflowAgents follow a predefined, deterministic execution path (like sequential or parallel steps) without LLM reasoning overhead for the flow itself.

Question 2

When should I use the Coordinator Pattern?

Accepted Answer

The Coordinator Pattern is best for complex problems that require different specialists. A root coordinator analyzes the user request and delegates sub-tasks to specialized agents (e.g., a 'security specialist' or 'frontend specialist') before synthesizing the final result.

Question 3

How do agents communicate with each other in this system?

Accepted Answer

Agents share information via session state. For example, one agent can save extracted data to the ToolContext state, and a subsequent agent in the workflow can retrieve that data for validation or storage.

Question 4

Does this skill support parallel execution?

Accepted Answer

Yes, it includes a ParallelAgent pattern specifically designed to run multiple independent agents concurrently, which is ideal for speeding up tasks like multi-stage analysis or independent validation checks.

AI Agent Orchestration

Características Principales

Casos de Uso

AI Agent Orchestration

Características Principales

Casos de Uso