Idea Generation

This is a structured workflow for generating research ideas with built-in reality checks. It walks you through generating 3-5 ideas, iteratively refining them for up to 5 rounds each, then running novelty checks against Semantic Scholar to see if your brilliant insight already exists in the literature. Each idea gets scored on interestingness, feasibility, and novelty, with the explicit goal of being conference-paper worthy. The included Python script automates the literature search part, which is honestly the most tedious bit of validating whether an idea is worth pursuing. It's opinionated about keeping ideas feasible (no massive datasets or compute) and not overfitting to specific benchmarks. Best used early in a research project when you're still figuring out what to work on.