Different Results in Playground Vs Python Client Library

I have an assistant whose only job is to figure out which other assistant to forward a message to. The other assistants each have their own functions and resources, so this assistant’s job is to figure out which one is the most appropriate to use based on the user’s input. The assistant is set to return json, the assistant’s instructions contain the term ‘json’, and when using the python library, the run instructions also include the term ‘json’. The other assistants all have similar set up and work just fine. But this assistant, when using the python client library just responds in a confused way as if it doesn’t have any of the instructions. When I inspect the assistant and run in python, everything looks right - I can see the instructions and everything on both objects. In the playground in the dashboard, the assistant works as expected. Anyone have a similar experience or have any solutions?

Open the thread that was created through your code and see what was sent. You can do this in the playground. There is almost certainly a discrepancy between your expectations and what’s being executed

Thank you for the reply, the method was helpful you were correct - I missed in the documentation that the run instructions override the assistant instructions and I was unintentionally doing that. Silly mistake on my part.

No problem. It happens to the best of us. Happy coding