skill

kaggle-dominator

Evidence-driven Kaggle strategy for coding agents: trustworthy validation, BEST_KNOWN protection, resource budgets, Pareto selection, and safe submission gates.

KapaSique/kaggle-dominatorcommunitylistedunreviewed

Use skill

Set up

Skills are installed by placing their folder where your agent loads skills.

Download ZIP from GitHub Open the repository

The download comes from github.com/KapaSique/kaggle-dominator itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 26, 2026

Kaggle Dominator

An evidence-driven strategy and execution skill for legitimate Kaggle competition work across tabular, computer vision, NLP/audio, simulation and agent ladders, code competitions, and judge-scored hackathons.

It is the competitive brain: live reconnaissance, validation design, BEST_KNOWN protection, experiment portfolios, ensembling, campaign focus, and submission decisions. Pair it with a separate kaggle infrastructure skill for authentication, downloads, kernels, and API plumbing.

What v3 adds

Universal routing across all major Kaggle formats.
Portfolio states (ACTIVE, MONITOR_ONLY, PAUSED, CLOSED) and explicit budgets for attention, compute, submissions, storage, and deadline slack.
A candidate queue with exploit/explore/audit lanes and stop/continue gates.
Fresh-score, metric-direction, submission-availability, and provenance checks.
Clear authority boundaries for submissions and public actions.
Compliance, attribution, reproducibility, and anti-leakage safeguards.
Learning checklists, public-resource guidance, an account scorecard, measured cross-domain lessons, and bounded autonomous loops.
Contract tests and Codex discovery metadata.

Core loop

live recon
  → front card and resource budget
  → reproduce strongest compliant baseline
  → trustworthy validation / arena
  → candidate batch with cheap gates
  → full measurement and diversity analysis
  → submission decision with authority + quota checks
  → fresh score, ledger update, next decision

A candidate never replaces BEST_KNOWN until the correct real metric confirms it. If validation disagrees with the leaderboard or ladder, repair validation before continuing to tune.

References

Need	Reference
Multiple fronts and resource focus	`campaign-control.md`
Cross-domain competitive process	`grandmaster-playbook.md`
Measured account history	`scorecard.md`
Tabular	`tabular.md`
CV, NLP, audio, signals	`deep-learning.md`
Simulation and agents	`simulation.md`
Code competitions and hackathons	`code-and-hackathon.md`
Autonomous bounded loops	`autonomous.md`
Self-curation	`self-improvement.md`
ML craft	`learning-craft.md`
Learning sources and attribution	`resources.md`
Winning solution patterns	`winning-solutions.md`
Tools and communities	`arsenal.md`

Install

Codex project-local installation:

git clone https://github.com/KapaSique/kaggle-dominator.git \
  .agents/skills/kaggle-dominator

Codex user installation:

git clone https://github.com/KapaSique/kaggle-dominator.git \
  ~/.agents/skills/kaggle-dominator

Invoke it explicitly as $kaggle-dominator, or let it trigger on competitive Kaggle strategy requests. It deliberately does not trigger for badge collection, account configuration, or CLI installation alone.

Verify

python3 -m unittest -v tests/test_skill_contract.py
python3 /path/to/skill-creator/scripts/quick_validate.py .
bash -n scripts/*.sh
python3 -m py_compile scripts/*.py tests/*.py

License

MIT