create-routine

To turn a manual task into a scheduled Python routine, you define the routine's purpose, choose between an AI-powered or systematic execution path, and generate the required Python script structure with Makefile integration for the scheduler.

AI routines use Claude Code agents for complex analysis and reporting which incurs token costs, whereas systematic routines execute deterministic Python scripts for tasks like API polling or file operations without token costs.

You schedule recurring API polling and file operations by creating a systematic Python routine, configuring the execution schedule, and registering the generated script within the scheduler to run automatically.

Yes, you can use Claude Code agents for periodic report generation by configuring an AI-powered routine, setting the desired schedule, and letting the automation framework handle the timed execution and logging.

Yes, the EvoNexus scheduler expects automated routines to include Makefile integration alongside the Python script structure to ensure deterministic execution, proper timeout handling, and logging for scheduled tasks.

Limitations include managing timeout configurations for long-running operations and ensuring deterministic Python execution for log cleanup, as complex unpredictable tasks may require AI routines instead of systematic scripts.