What problem does it solve? Researchers risk wasting effort on ideas that have already been published or closely resemble existing work. This Skill performs a rigorous, multi-source novelty check on a research idea or method before you commit to it, producing a scored report with the closest prior work and a clear proceed/modify/abandon recommendation. ## Core Features & Use Cases - Multi-Source Search: Queries WebSearch, Semantic Scholar, DeepXiv, recent arXiv preprints, and your internal wiki (papers, concepts, ideas, claims) in parallel to find similar work. - Review LLM Cross-Verify: Submits the method signature and top similar works to an independent review model for an unbiased novelty judgment, with a conservative scoring rule that takes the lower of the two scores. - Anti-Repetition Checks: Scans failed and in-progress ideas in the wiki to prevent repeating known dead ends or duplicating ongoing internal work. - Use Case: Before starting a project on sparse LoRA for edge devices, run the novelty check to receive a 1-5 novelty score, the top 3-5 closest papers with differentiation points, and a recommendation on whether to proceed, modify, or abandon the idea. ## Quick Start Ask the assistant to run a novelty check on your research idea, for example by providing a short description of the method or the slug of an existing idea page in the wiki.
