FrankS-IntelLabFrankS-IntelLabCommunityยท1 Agent Skills Included

agentic-kaggle-skill

End-to-end Kaggle competition workflow through scored submission

Runs complete Kaggle competition workflows from rule analysis and fold design to modeling, ensembling, and scored submission. Eliminates guesswork around validation splits, hidden-test failures, OOM errors, and vague scoring errors with structured debugging loops. Offloads heavy training to Kaggle notebooks and manages multi-stage producer-consumer pipelines with private artifact datasets. Tracks every run's CV score, submission receipt, and reproducibility artifacts until a public score is reco
npx skills add FrankS-IntelLab/agentic-kaggle-skill --all -g -y

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install agentic-kaggle-skill?โ–ผ

Run `npx skills add FrankS-IntelLab/agentic-kaggle-skill --all -g -y` in your terminal to install the skill globally.

How to automate a Kaggle competition workflow?โ–ผ

This skill guides your agent through the full loop: reading competition rules, building metric-correct baselines, designing folds, iterating with OOF validation, and submitting for a scored result.

How to debug Kaggle code competition failures?โ–ผ

It provides a structured debugging loop that classifies failures like OOMs, timeouts, path mismatches, and schema errors, then patches and resubmits until the submission is scored.

Does it work with Codex and other agents?โ–ผ

Yes. It follows the open SKILL.md standard with first-class Codex support via agents/openai.yaml, and also works with Hermes and other compatible agents.

Can it handle heavy GPU training for Kaggle?โ–ผ

Yes. It offloads heavy training and inference to Kaggle notebooks, splits work into producer and consumer pipelines, and stores model artifacts as private Kaggle datasets.

Related Repositories in Data & Analytics

View All in Data & Analyticsโ†’