task-runner

You can automate todo task execution by using a task-runner that reads todo files, validates context, delegates work to a task-executor agent, and moves completed items to done with automatic results reporting.

Sentinel-based control is a mechanism that initializes a task loop to continuously process todo tasks, automatically stopping execution when all tasks are completed or when the sentinel signals a halt.

To set up a continuous loop, create the initial sentinel and start the task loop, which will automatically load tasks, validate context completeness, delegate execution, and retry on failures until finished.

Yes, the task-runner enforces completeness by checking each todo task file for required sections and automatically augmenting missing parts before delegating the work to the task-executor agent.

When task delegation fails, the task orchestration loop automatically retries the execution, ensuring hands-free workflow progression until the task is successfully completed or the sentinel stops the loop.

Yes, you can use this task orchestration approach without external dependencies, as it operates entirely within Claude-based workflows by reading local todo files and delegating to an internal task-executor agent.