After you've run deep research queries and accumulated findings in memory, this pulls everything together into a structured report with evidence grading and contradiction resolution. It searches across research namespaces, scores each finding as high/medium/low confidence based on source quality and agreement, flags conflicts between claims, and uses neural prediction to rank relevance. The output is a formatted synthesis with executive summary, ranked findings, limitations, and actionable recommendations. It's the capstone step that turns a pile of research fragments into something you can actually hand off or act on. Stores the final report back to memory under research-synthesis namespace.
npx skills add https://github.com/ruvnet/ruflo --skill research-synthesize