SequentialAgents, Autonomous Execution

[a-ml]

Hello,

Is it possible to run Sequential Agents or Agents without user question? The agent should starting working immediately on the task as soon as it starts, or implementing a sleep function.

[Jacksunwei]

Could you elaborate? Maybe you could provide a concrete example or a minimal agent snippet?

[a-ml]

What i meant is.

I type adk run nameoftheproject on cli and I receive the information below:

Which is waiting for a User input.

What I want is that as soon as I type adk run ... it starts working following the instructions and remove the need for user/human input, or add it only where is needed.

[zbigniev]

@a-ml I believe you are referring to adk cli or web, where the first input is expected from the user, whereas what you want is for the sequential agent to start running without user input https://github.com/google/adk-python/blob/main/src/google/adk/cli/cli.py#L85.

I think cli and web have those limitations. You'd have to implement a custom runner and session and run them in a while True loop. Have a look at the tutorial - Agent Team (https://google.github.io/adk-docs/tutorials/agent-team/) which creates a custom runner and session