My agent is not auto advancing to next steps even when max steps is at 5.
And it seems that my agents memory is lagged by 1.
Please see image:
Specifically,
I’ve noticed that when I send my very first message:
I ate a chicken chop and I walked for 2 km the agent replies with its generic greeting
(“Hello there! How’s your day going?…”), and only after I send a second message does it trigger the addFood and addExercise tool calls.
(“Hello there! How’s your day going?…”), and only after I send a second message does it trigger the addFood and addExercise tool calls.
What’s happening:
1st user message → generic greeting only
2nd user message (which is just a
.
.
) → adds food/exercise and shows summary
What I expected:
1st user message → immediately estimate calories/macros and log the exercise
I’ve attached a screenshot of the behavior. Could you point me to the specific prompt tweak or model config (function_call mode, streaming, max_steps, etc?) that will make the agent process the very first user input end‑to‑end?
Thanks!
NOTE I am not using a workflow, just tools, but this is a simple use case.