What problem does it solve?
This Skill eliminates the guesswork and hours of wasted effort when debugging runtime failures, tracebacks, unexpected behavior, value/config mismatches, or tensor/data shape issues, helping you pinpoint the exact root cause without modifying any project code.
Core Features & Use Cases
- Systematic Backward Tracing: Start at the exact failure or observation site and walk backward through callers to find where the incorrect value, config, or input was first created.
- Structured Variable Ledger: Track important variables' origins, transformations, concrete values/shapes, and next consumers to avoid missing critical context during investigation.
- ML/PyTorch Specialized Guidance: Includes targeted checks for tensor shape tracking through dataset → transform → collate → model → loss pipelines, train/eval mode mismatches, dtype/device issues, and common convolution/pooling output calculations.
- Use Case: If you encounter a tensor shape mismatch during PyTorch model training, use this Skill to trace the shape through each pipeline step to isolate the first point of divergence between working and failing runs.
Quick Start
Use the codepath-debugging skill to trace the root cause of the latest traceback you encountered in your Python project without modifying any project files.
