Agno v1.8.5 (I don’t want to upgrade to v2 as of now since it will break my project)
This is what I want the logger to do
Let’s say there is an Orchestrator, a team, under which there are 2 teams (each consisting of 3 agents) and 2 other agents. Each of these agents have at-least say 3 tools, either user-defined or Agno tools. And all these entities use OpenAIChat 4.1-mini as the model (again, model may vary, not the point here). What I want to do is, from the moment the user gives the query till the last step, everything should be logged into a CSV file. At any point of time, the metrics and tool call done and agent invoked should be accessible so that the corresponding data can be logged into CSV.
This is actually important for me now, since my existing architecture somehow exhausts my OpenAI credits and I need to see what is draining the credits.
Also, is there any way I can make the team do only a single call to an agent and not repeated calls?
Hey @TheMaster24, Thanks for reaching out and supporting Agno. I’ve shared this with the team and we’re working through all requests one by one and will get back to you soon. If it’s urgent, please let us know. We appreciate your patience!
If the goal here is to review the usage of the Agent/Team, I’d recommend using storage with your Team and going through the stored session to review all the token usage.
Also, is there any way I can make the team do only a single call to an agent and not repeated calls?
Can you please share more here? Is the concern to limit parallel tool calls or is the Team making unwanted tool calls?
Thanks @yash for the suggestion. I will check out using storage, but then can a Team have storage + memories as well or is it only specific to Agent only?
The 2nd part was related to the Team making continuous agent calls like, say my Team has 3 agents. First run, it calls all 3 agents, sequentially. Then it takes time to process the data or something (around a min) and then it again calls the 3 agents sequentially with almost the same query it has used previously to call (with debug_level=2, I saw some queries that the Team sent to the Agent to make). So it is more like, redundant calls, wasting my time.