dag-execution-tracer

This skill provides an automated bug-fixing workflow that performs root cause analysis, generates fixes with Codex, and validates changes with testing. It's designed for developers to systematically debug issues by combining multi-model reasoning with iterative validation. Use it when you need a structured, AI-assisted approach to diagnose and resolve complex bugs.

This skill helps developers diagnose and fix common machine learning training issues like loss divergence, overfitting, and slow convergence. It provides systematic debugging to identify root causes and generate fixes for training problems. Use it when you encounter poor validation performance or training instability to restore model convergence.

This skill provides a structured four-phase debugging framework to replace random code changes with systematic problem diagnosis. It helps developers methodically investigate bugs, errors, and unexpected behavior by forming specific hypotheses and testing single changes. Use it when under time pressure or when quick fixes seem obvious to ensure reliable problem resolution.

This skill systematically traces bugs backward through the call stack to identify their original triggers rather than just fixing symptoms. It's designed for use when errors occur deep in execution with unclear data origins or long call chains. The approach involves observing symptoms, finding immediate causes, and repeatedly asking "what called this" until reaching the source.