Manual and Mandatory Tool Call

These two blocks allow you to perform highly controlled calling of tools and storing of results.

The Mandatory Tool Call block uses a LLM to craft the arguments to pass to a tool, but force the LLM to call exactly this tool, exactly once.

The Manual Tool Call lets you entirely write the arguments to pass to the tool, using CEL expressions

You can then decide what to do with the output of the tool call:

• Store it into the state

• Store it into the scratchpad

• Add it to the conversation history, so that it can be picked up by subsequent Core Loop or LLM Request blocks