Run prompt
Large language models (LLMs) are a powerful tool that can be used in many different ways. With the Run prompt node, you can run a simple LLM prompt right in your Flow.
The LLM Flows skill adds this Flow node to your Flow designer.
How to use?
The Run prompt node can be used anywhere in your Flow. It sends a prompt to the LLM, and receives a response. That response is then stored in a variable, which can in turn be used in subsequent parts of your Flow.
In the above Flow, the user is asked to name a colour. The prompt sent to the LLM asks it to create a rhyme about that colour. The response containing the rhyme is stored in the llm_colour_response variable, which is used in the next Say node to repeat it to the user:
Node properties
Assign to
The variable that the LLM's response should be stored in
Include transcript
The request to the LLM will contain a transcript of the conversation so far
This can be helpful when asking the LLM to say something about the conversation, e.g. to create a summary or filter out specific information
Resources
Last updated
Was this helpful?