Dex Async Task Manager

To manage long-running async tasks without blocking your agent, you can decouple agentic thought from background execution, allowing processes like model training or batch processing to run outside the main loop.

The best way to run background tasks for batch processing is to initiate the process asynchronously, which allows the system to track task status and capture results without waiting for completion.

Your agent blocks when executing external API calls exceeding 10 seconds because the process runs within a single turn, a limitation solved by decoupling the execution into an asynchronous background task.

Yes, you can retrieve the results of a background process after it completes by listing and retrieving detailed task logs, which capture the final status and output of the asynchronous execution.

You start a python script as a background process by passing the command, such as 'python scripts/long_running_job.py', to the task manager, which initiates it asynchronously and tracks its execution.