What problem does it solve?
It helps you assess whether a research idea or method is genuinely novel by comparing it against existing work found on the web, in Semantic Scholar, and inside the EmpiricalWiki graph, then producing a conservatively scored novelty report with actionable next steps.
Core Features & Use Cases
- Multi-source novelty verification: Performs WebSearch + Semantic Scholar + wiki-internal checks to find closely related prior art and failed/redundant ideas.
- Method-signature extraction: Accepts free text, a wiki idea slug, or an arXiv URL/title and extracts a “method signature” plus search keywords to drive targeted retrieval.
- Optional cross-validated review step: Uses a Review LLM to independently score novelty (1-5) and to recommend proceed/modify/abandon; a quick mode can skip the review step.
- Read-only safety: Uses wiki data strictly for querying and does not modify any wiki content.
Quick Start
Use the novelty skill to evaluate an idea by running: I want to check the novelty of the method described as "..." in quick mode.