kaggle-kernel-runner
14790897/my-skills
Push, monitor, debug, and iterate on Kaggle kernel notebooks using the kaggle CLI. Covers: nbformat notebook generation, kernel-metadata.json configuration (GPU/dataset), push→poll→debug→fix→repush cycle, papermill timeout avoidance, and output download. Use when the user wants to run a notebook on Kaggle cloud GPUs and iteratively fix errors until it succeeds.
kaggle-notebook-format
14790897/my-skills
Use this skill when creating Python scripts that target Kaggle Notebooks. This skill ensures correct cell separator format (# %% [code] and # %% [markdown]), proper markdown cell formatting (each line prefixed with #, no triple quotes), and Kaggle-specific conventions (Chinese comments, English output, %pip install). Also covers common pitfalls: OpenCV video decoding limits and Chinese text rendering with PIL.