debug-loop-run

Debug long-horizon AI agent loops by automatically synthesizing execution history to identify failure patterns and generating actionable issues to prevent recurring stalls or crashes. The skill scans event logs for action failures, SIGKILLs, and evaluation errors to pinpoint root causes.

Signal classification in AI agent loop debugging categorizes identified issues into Fault (BUG) and Effectiveness (ENH) signals, providing a clear summary of loop health. This automated state inspection helps distinguish between critical errors and enhancement opportunities within the execution history.

Track recurring failures by automatically creating and staging structured issue files from synthesized execution logs. This issue management process prevents duplicate reports and streamlines the development workflow by organizing identified faults and effectiveness signals.

Yes, the debug-loop-run skill works within the little-loops ecosystem, satisfying requirements for automated state inspection, signal classification, and issue tracking. It is designed for long-horizon AI software development workflows requiring durability and verification.

The best way to analyze loop execution history for failure patterns is to automate the synthesis of event logs, classifying issues into actionable BUG and ENH signals. This approach identifies root causes of loop termination and stages structured issue files for tracking.