What is the difference between developer UI (adk web) and terminal (adk run)?

[tvcf-soony]

※ (Korean novice developer) Since it's not an English speaking country, the questions in English can be a bit awkward. Hope you understood the questions.

• I just started Google ADK learning to realize the ideas I want with AI agents.
• When you develop and test an agent with google adk, you usually use the "adk web" command to conduct an agent function test with the developer UI.
• However, among the ways to interact with agents in ADK, there is also an "adk run" method.
• Create an agent with the default project structure (agent_folder/below __init___py, agent.py , .env), implement the Agent (LLMAgent) on agent.py , and assign it to the root_agent variable.

My question here is, when you test with the adk web UI, you can check Sessions, Events, States, and so on, and create conversation sessions on your own.

1. What happens when you run with "adk web"?
2. The difference between "adk web" and "adk run" is not just the difference between testable or not with a simple UI, what's the difference?
3. If I implement the default agent and don't set up a session with the runner separately, would it be impossible to test the agent in Terminal (adkrun)?

(I have a lot of difficulty learning due to lack of references related to Google ADK in Korea. Your help would be appreciated!)