ideate

To generate research ideas, this Skill uses a five-phase pipeline that scans the wiki landscape, brainstorms using dual models, filters results, validates deeply, and writes final ideas to wiki pages. It connects established wiki gaps and weak claims to external literature for traceable novelty.

The best way to automate research ideation is running the pipeline with the --auto flag, which enables fully automatic exploration from landscape scanning to wiki writing. You can also use --max-ideas to control output volume and --skip-validation to increase processing speed.

Yes, the ideation pipeline is specifically tailored for in-silico social science topics. It leverages wiki gaps, weak claims, and external literature to produce research ideas that are both novel and actionable for social science research teams.

Wiki-grounded brainstorming validates ideas by linking them to relevant papers, concepts, and claims within the knowledge base during the deep validation phase. This ensures every generated idea is traceable to established literature and identified wiki gaps before being written.

A limitation of automated ideation is that fully automatic runs using --auto bypass interactive review, which may reduce human oversight. Additionally, skipping the validation phase with --skip-validation trades output quality for speed, potentially producing less grounded research ideas.

Yes, you need a living wiki knowledge base because the generator anchors its end-to-end ideation pipeline to existing wiki pages, graph edges, and identified knowledge gaps. Generated ideas are written directly to wiki/ideas directories and associated graph edges.