Solver kill-switch and progress monitoring for schedule generation. Use when aborting runaway solvers, monitoring long-running schedule generation, or integrating abort checks into solver loops.
The solver-control skill provides operational control for schedule generation solvers, enabling graceful interruption of runaway jobs and real-time progress tracking. This skill is designed to address the challenges faced when solvers take too long to complete, allowing for immediate action through kill-switch functionality. It can be used to monitor the progress of long-running tasks or to integrate abort checks within solver loops, providing an efficient way to ensure that processes do not run indefinitely and consume resources unnecessarily. Furthermore, the skill offers mechanisms for debugging solver performance and terminating jobs in emergency situations where manual intervention is needed.
You can abort a running solver by sending a POST request to the `/scheduler/runs/{run_id}/abort` API endpoint, providing a reason for the abort and the user who requested it.
Yes, but any force kill operations in production systems require human approval, as abrupt termination may result in data corruption.
The skill requires a database context and does not need a git context. It can also operate without a heavy compression setting for context.
Yes, the solver-control skill can work in parallel with the resilience-dashboard and schedule-validator, but must be serialized with safe-schedule-generation and SCHEDULING.
In emergency situations, you can directly set the abort flag in Redis to trigger an immediate stop of the solver without relying on the API.
Quick Setup:
.claude/skills/