This node supplies an LLM agent with a set of tools and the conversation history, and prompts it with the user-provided query.
The conversation table is expected to have at least two string columns that define previous conversation. If this is the start of the conversation, the conversation table can be empty.
The agent always receives the full conversation table as context, which can lead to slower execution times for longer conversations, or execution failures if the context becomes too large for the LLM. If you experience such issues, you can truncate the conversation table by only keeping the last few messages, or use an LLM to summarize the conversation held so far.
Note : If you use the Credentials Configuration node and do not select the "Save password in configuration (weakly encrypted)" option for passing the API key for the LLM, the Credentials Configuration node will need to be reconfigured upon reopening the workflow, as the credentials flow variable was not saved and will therefore not be available to downstream nodes.