Emit Output

Output of the Agent to the user is primarily craeted by the Core Loop and llm-request blocks.

In both of these blocks, the LLM ultimately generates text that can be sent to the user.

However, you may want more control on what exactly the Agent responds to the user.

For that, you can use the Emit Output block.

The Emit Output block contains a template </agents/structured-visual-agents/expressions, either a CEL template or a Jinja template, that is expanded and evaluated using the current state and scratchpad. This can be used for example:

• At the end of the turn to provide an answer with predictible formatting

• Before a long operation, to indicate to the user that this will take some time.