[FEATURE] Iterative implementation in tasks due to inconsistent delegation when working with smaller LLM
Feature Area
Task management
Is your feature request related to a an existing bug? Please link it here.
I tried to implement a crew that contains a data analysis agent and an SQL data fetching agent, with the SQL data fetching agent having the tool to connect to my database. The idea was to let the data analysis agent first look at initial data and spot any anomalies, suspect a root cause, and then iteratively investigate for the root cause through request more data from SQL data fetching agent and analysis the newly fetched data. I built it as hierarchical process but encountered problem when manager agent keep delegating wrong task to the wrong agent, or just pass repeated task to the same agent. My thought is to implement with two task, data fetching and data analyzing, with conditional task, but it seems that conditional task only check sequentially, meaning if I do data analyze -> need more data -> fulfill conditional task condition -> fetch more data -> can't return back to data analyze task.
Describe the solution you'd like
Implementation of iterating tasks such as routing for tasks, or spawn tasks sets: x amount of tasks combination based on config
Describe alternatives you've considered
No response
Additional context
No response
Willingness to Contribute
I can test the feature once it's implemented
So , I also have. small doubt in this process , are you able to send the result from one task to another as it is important in this situation or are you logging them into some file and checking from that?
Faced similar issues , where allow_delegation is not working as intended and waste of so much tokens.
This issue is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.
This issue was closed because it has been stalled for 5 days with no activity.