Retrieving Agent Session State Without Triggering Execution in Agno

[YZx0pa]

Hi,
I’d like to check if there’s a recommended way to fetch the session_state from an agent more flexibly.

In my setup, I’m running the agent inside a FastAPI app. The agent is instantiated when the server starts, and multiple users interact with it via API calls. For each request, I call agent.run(user_input, user_id=...), which may trigger function calls and update dynamic data into session_state (e.g., a dictionary collecting information throughout the chat).

Currently, it seems I can only access the correct session_state after calling agent.run(...). However, I’d like to know if there’s a way to directly retrieve(or even update the session_state) by the user_id or session_id, similar to how agent.get_messages_for_session() works — without needing to run the agent first.

Could you please advise if there's a cleaner or more flexible way to handle this in Agno?

Thanks!

[YZx0pa]

Additionally, I’d like to store user-specific credential information inside the agent’s session for use during function calls. Since these credentials are sensitive, I plan to set add_state_in_messages=False to prevent them from being sent to the LLM. In my function, I would access this session-level state as shown below:

def my_function_tool(agent: object, data: dict):
# Access session_state here

I’d like to confirm the following:

1. Can I update the session_state for a specific user_id before calling agent.run()?
   This is important because I want to inject credential info into the session so that it can be used in the function.

2. Does handling function calls this way, with add_state_in_messages=False, guarantee that the session_state is excluded from any prompts or messages sent to the LLM?
   I want to ensure credentials or internal data remain fully isolated from the model.

Alternatively, is there a more elegant or recommended way in Agno to securely handle user-specific credential information when triggering a function call through the agent?
Thanks!

[monali7-d]

Hey @YZx0pa
So sorry for the delay in getting back to you — this slipped through the cracks and we’re now working through all the requests one by one. Thank you so much for reaching out and supporting Agno! I’ve shared your message with the team, and we’ll get back to you as soon as we can. If this is urgent, please feel free to let us know. Really appreciate your patience!