The agent keeps a local database on the workcell machine, and on rare occasions that file can develop internal inconsistencies that stop scheduled jobs from running.
A scheduled check now inspects database integrity and records what it finds. If a recognized, safely repairable problem is detected, the agent fixes it automatically the next time it starts, and logs the outcome either way. A repair that does not succeed no longer prevents the app from starting.