Novelty Assessment

This runs a systematic literature search loop to figure out if your research idea already exists. It does up to 10 rounds of searches across Semantic Scholar and arXiv, checking each batch of papers for overlap. The harsh critic persona is useful here: it won't let you ship a trivial extension as novel work. You get a binary novel or not novel decision with confidence level and specific papers that come close. Minimum three search rounds before it'll declare something novel. Best used before you commit weeks to a direction, and the structured JSON output feeds nicely into related work sections if you proceed.

Install to Claude Code

npx -y skills add lingzhi227/agent-research-skills --skill novelty-assessment --agent claude-code