dsh-stall-guard

During automated task execution, processes often stall silently due to issues like hung tool calls or infinite loops. Traditional monitoring tools may forcefully terminate these tasks, leading to workflow disruption and context loss. The dsh-stall-guard plugin is designed to address this critical problem as a dedicated task watchdog.

Its core capability lies in intelligent monitoring and ladder-style guidance. The plugin continuously analyzes event streams and operational statuses to accurately distinguish between "true stalls" and "long-running processes," preventing false alarms. When a task is genuinely stuck, instead of termination, it provides progressive prompts—from diagnostics to fixes, and finally to redirects—to guide the task through self-recovery, ensuring workflow continuity within a safe boundary.

This plugin is ideal for developers and researchers who run complex, long-duration tasks such as large-scale data processing, extensive text generation, or continuous integration pipelines. It enhances automation reliability, reduces the burden of manual intervention, and safeguards task integrity by never unexpectedly terminating operations.