Back to catalog

Agentic Kaggle skill

A Kaggle workflow skill for autonomous competition iteration, notebook replication, submission debugging, and score monitoring.

Repository
FrankS-IntelLab/agentic-kaggle-skill
Skill path
SKILL.md
Systems
Claude Code, Codex, Cursor
Install
curl -sL https://raw.githubusercontent.com/FrankS-IntelLab/agentic-kaggle-skill/main/SKILL.md -o ~/.hermes/skills/data-science/agentic-kaggle/SKILL.md
GitHub stars
97
Updated
2026-05-13
Source
README · SKILL.md

What it is

agentic-kaggle-skill is a data science competition skill. Its README positions agents as Kaggle teammates that can research top notebooks, debug submissions, iterate on improvements, and monitor kernels or scores.

Why it is useful

Kaggle work is repetitive and evidence-driven. A focused skill can help an agent follow competition workflow instead of producing generic data science advice.

Best uses

  • Replicate a top Kaggle notebook with dependencies.
  • Diagnose submission format errors and failed kernels.
  • Plan agent-assisted competition iterations.

Notes

This is useful for competition workflow, not a promise of winning. Keep leaderboard rules, data leakage constraints, and competition terms explicit.