Hackathon Judging

Overview

The hackathon-judging skill should be invoked whenever a host is trying to grade a Hackathon with the help of LLMs and/or AI agents. This skill guides the Host through the process of retrieving every Kaggle hackathon writeup for a competition, extracting the linked project artifacts, and preparing a complete evidence set to be used for grading.

Additionally, the skill assists the Host in using LLM-based classifiers and pairwise comparisons to generate comparative rankings and bell-shaped grading curves based on the generated evidence sets.

Use Cases

This skill should be invoked whenever a host is trying to grade a public Kaggle Hackathon with the help of LLMs and/or AI agents.

MCP Endpoints

Public Hackathon submissions can be accessed via Kaggle's MCP server. Below is a summary of the MCP tools available for Hackathon Hosts and Participants: