improve-loop-system

To prevent autonomous agent loop failures, you analyze execution logs and verification evidence to systematically refine project rules and sensors. This approach transforms operational data into structured improvements, directly addressing memory loss and cost overruns.

The best way to detect system drift is by scanning execution data for stale documentation, unused sensors, and architectural deviations. This maintains system health by identifying bottlenecks and verifying that continuous operations match project rules.

To update linting rules and safety boundaries, you aggregate journal logs and verification failures to categorize issues into actionable improvements. This structured evolution creates new boundaries that prevent recurring errors in autonomous loops.

Evidence-based rule evolution works for continuous quality assurance by aggregating cost metrics and failure logs to systematically improve sensors. It satisfies requirements for regression testing and automated system auditing in long-running projects.

You should not use automated system auditing for agent loop engineering when your project lacks operational evidence like failure logs or cost metrics. Without execution data to analyze, the skill cannot identify systemic bottlenecks or suggest structured improvements.