JosiahSiegel/kernel-metadata
plugins/kaggle-master/skills/kernel-metadata/SKILL.md
Kaggle `kernel-metadata.json` setup and repair. PROACTIVELY activate for: (1) creating kernel-metadata.json, (2) fixing metadata validation errors, (3) setting `id`, `title`, `code_file`, language, or kernel type, (4) configuring private/public, internet, GPU, or accelerator behavior, (5) attaching dataset_sources, competition_sources, kernel_sources, or model_sources, (6) preparing metadata before `kaggle kernels push`, (7) converting script/notebook metadata. Provides: schema checklist, valid fields, source arrays, defaults, and push-readiness review.
$ install --global
skillsauthnpx skillsauth add JosiahSiegel/claude-plugin-marketplace kernel-metadataInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: May 28, 2026, 7:33 AM104.6s1 file scanned
SKILL.md
- name:
- kernel-metadata
- description:
- |
- Kaggle `kernel-metadata.json` setup and repair. PROACTIVELY activate for:
- (1) creating kernel-metadata.json, (2) fixing metadata validation errors, (3) setting `id`, `title`, `code_file`, language, or kernel type, (4) configuring private/public, internet, GPU, or accelerator behavior, (5) attaching dataset_sources, competition_sources, kernel_sources, or model_sources, (6) preparing metadata before `kaggle kernels push`, (7) converting script/notebook metadata. Provides: schema checklist, valid fields, source arrays, defaults, and push-readiness review.
Kernel Metadata
Overview
Use this skill to create, validate, or repair kernel-metadata.json for Kaggle notebook/kernel pushes. Keep metadata minimal, explicit, and aligned with the local code file.
Required and Common Fields
| Field | Purpose |
|---|---|
| id | Full kernel id, usually <owner>/<slug> |
| id_no | Numeric id when available |
| title | Human-readable notebook title |
| code_file | Local source file to run |
| language | python, r, or rmarkdown |
| kernel_type | script or notebook |
| is_private | Defaults to true |
| enable_gpu | Defaults to false |
| enable_internet | Defaults to false |
| dataset_sources | Attached dataset references |
| competition_sources | Attached competition references |
| kernel_sources | Attached notebook/kernel references |
| model_sources | Attached model references |
Validation Procedure
- Confirm
kernel-metadata.jsonis valid JSON. - Confirm
iduses the expected owner/slug and matches the intended Kaggle notebook. - Confirm
code_fileexists in the same folder passed tokaggle kernels push -p <folder>. - Confirm
languageis one ofpython,r, orrmarkdown. - Confirm
kernel_typematches the file and execution mode:notebookfor.ipynb,scriptfor scripts. - Confirm privacy and internet settings are deliberate. Default private and no internet are safer.
- Confirm source arrays contain valid public Kaggle references and no local filesystem paths.
- Load
notebook-lifecyclebefore push or pull workflows.
Source Attachments
Use dataset_sources, competition_sources, kernel_sources, and model_sources to attach Kaggle resources. Prefer exact owner/slug references where available. For competition notebooks, ensure the competition source is attached and rules allow all data/model sources used.
id and Browser URL Construction
The id field identifies the Kaggle notebook/kernel using the documented suffix format owner/kernel-slug. When id is present and valid, agents may derive the human browser URL:
https://www.kaggle.com/code/<owner>/<kernel-slug>
Validate before deriving: id contains exactly one /, both owner and slug are non-empty, id is not a full URL, and the slug comes from the actual id rather than a generated title guess. If id_no is also present, treat it as a stable numeric identifier for API operations where supported, but do not use id_no alone to construct the browser URL.
The derived URL is a practical browser path based on the documented identifier suffix, not a separately documented CLI return value. Private notebooks may require sign-in or permissions, and renames or slug changes can make older derived URLs stale.
Accelerators and Metadata
enable_gpu is a legacy/simple boolean for GPU enablement, while CLI push can specify accelerator IDs with --accelerator <ID>. Valid accelerator IDs include NvidiaTeslaP100, NvidiaTeslaT4, NvidiaTeslaT4Highmem, NvidiaTeslaA100, NvidiaL4, NvidiaL4X1, NvidiaH100, NvidiaRtxPro6000, TpuV38, Tpu1VmV38, TpuV5E8, and TpuV6E8. Warn before high-end accelerators or long timeouts because they can consume quotas.
Safe Defaults
Keep is_private: true unless the user explicitly requests public visibility. Keep enable_internet: false for competition notebooks unless rules and reproducibility needs allow internet. Avoid embedding secrets in metadata or notebook source; recommend Kaggle Secrets for sensitive values.
Unsupported Metadata Fields
Kaggle kernel-metadata.json does not support secrets or arbitrary environment variables. Do not add fake fields such as secrets, environment, or env. For internet/GPU tunnel workloads, verify supported fields instead:
{
"enable_gpu": true,
"enable_internet": true
}
Secrets must not be modeled as metadata. If a CLI-pushed committed run needs auth-free connectivity, redesign the workflow to avoid secrets.
Repair Patterns
- Missing
code_file: identify the intended.ipynb,.py,.R, or.Rmdfile and set the field explicitly. - Wrong language: infer from extension only after confirming with the user when ambiguous.
- Broken source arrays: remove local paths and replace with Kaggle dataset, competition, kernel, or model references.
- Visibility changes: require explicit confirmation before changing private to public.
Sources
- Kaggle kernel metadata docs: https://github.com/Kaggle/kaggle-api/blob/main/docs/kernels_metadata.md
- Kaggle kernels CLI: https://github.com/Kaggle/kaggle-api/blob/main/docs/kernels.md
Related Skills
JosiahSiegel/ml-training
development
VerifiedTrustedCommunity
This skill should be used when the user asks to train, debug, scale, or improve ML models. PROACTIVELY activate for: (1) PyTorch, TensorFlow/Keras, JAX, Flax, Hugging Face Trainer/Accelerate training loops, (2) distributed training, DDP/FSDP/DeepSpeed, TPU/GPU setup, (3) mixed precision AMP/bf16, gradient accumulation, checkpointing, seeding, (4) overfitting, imbalance, loss functions, regularization, LR schedules, warmup, (5) memory optimization, gradient checkpointing, offloading, quantization-aware training. Provides: reproducible training best practices across deep learning and classical ML.
38SKILL.mdUpdated May 28, 2026
JosiahSiegel/ml-mlops
development
VerifiedTrustedCommunity
This skill should be used when the user asks to productionize, track, version, govern, monitor, or automate ML systems. PROACTIVELY activate for: (1) MLflow, Weights & Biases, Neptune, Comet, ClearML experiment tracking, (2) model registry, model versioning, artifact lineage, reproducibility, (3) Kubeflow, SageMaker Pipelines, Vertex AI Pipelines, Azure ML pipelines, Databricks workflows, (4) CI/CD, continuous training/evaluation, A/B tests, canary/shadow deployments, (5) drift detection, model monitoring, data validation, responsible AI governance. Provides: end-to-end MLOps architecture and operational safeguards.
38SKILL.mdUpdated May 28, 2026
JosiahSiegel/ml-inference-optimization
development
VerifiedTrustedCommunity
This skill should be used when the user asks to optimize, export, serve, compress, or accelerate ML inference. PROACTIVELY activate for: (1) latency, throughput, p95/p99, batching, concurrency, KV cache, memory, or cost issues, (2) quantization INT8/INT4, GPTQ, AWQ, bitsandbytes, pruning, sparsity, distillation, (3) ONNX export, ONNX Runtime, TensorRT, TorchScript, torch.compile, XLA, OpenVINO, Core ML, TFLite, (4) Triton, TorchServe, TF Serving, BentoML, Seldon, KServe configuration, (5) edge deployment, CPU/GPU/TPU/Inferentia serving. Provides: hardware-aware inference optimization and safe benchmarking.
38SKILL.mdUpdated May 28, 2026
JosiahSiegel/ml-hyperparameter-tuning
testing
VerifiedTrustedCommunity
This skill should be used when the user asks to tune hyperparameters, run sweeps, optimize search spaces, or use AutoML. PROACTIVELY activate for: (1) Optuna, Ray Tune, FLAML, AutoGluon, Hyperopt, Nevergrad, KerasTuner, W&B sweeps, (2) grid search, random search, Bayesian optimization, TPE, Gaussian processes, evolutionary search, (3) ASHA, Hyperband, successive halving, multi-fidelity optimization, population-based training, (4) learning-rate finder, batch-size search, early stopping, pruning, (5) reproducible sweep design and experiment analysis. Provides: budget-aware hyperparameter search strategy.
38SKILL.mdUpdated May 28, 2026