/ideate

To generate novel research ideas and avoid duplicating existing literature, this skill scans a wiki knowledge base and performs external literature discovery to identify gaps, producing ranked and feasibility-checked idea candidates for specified directions.

Validating research ideation using a knowledge graph involves applying multi-phase generation with dual-model brainstorming, deep novelty and review validation calls, and writing proposed and eliminated ideas back into the wiki to prevent repetition.

To run a literature search pipeline for pilot experiments, you execute the ideation pipeline with your research direction, allowing it to generate validated candidates and run lightweight pilot experiments that update idea pages with pass or fail outcomes.

Research ideation works with wiki automation by writing proposed and eliminated ideas into the wiki with specific failure reasons, rebuilding derived graph context to maintain accurate gap addressing and inspiration provenance for future searches.

Multi-model generation can brainstorm structured research paths by using independent models to force ideas through explicit generation paths like landscape-driven, incremental fixes, combinations, assumption breaks, or cross-domain transfer.

You can auto-generate a specified number of research ideas for a scientific topic by calling the skill with your research direction and setting the max-ideas argument, producing ranked candidates with optional pilot experiments.