sn-research-synthesis

Generate a decision-ready synthesis.md with evidence-strength judgments from sub-reports.

2|Updated May 19, 2026
One-click install
npx skills add https://github.com/aiyinluya/SenseNova-Skills-Studio --skill sn-research-synthesis-aiyinluya
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: sn-research-synthesis
Source: https://github.com/aiyinluya/SenseNova-Skills-Studio/tree/main/skills/sn-research-synthesis
Command: npx skills add https://github.com/aiyinluya/SenseNova-Skills-Studio --skill sn-research-synthesis-aiyinluya

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you produce a rigorous decision-oriented synthesis after multiple research sub-reports are completed, turning scattered evidence into an actionable judgment layer.

Core Features & Use Cases

  • Main-line judgments (2–5): Extracts the key reasoning threads that can support the final write-up rather than duplicating sub-report content.
  • Evidence strength labeling: Assigns high/medium/low confidence to each main judgment with justification, including conditions for applicability.
  • Cross-dimension consensus & conflict resolution: Summarizes what multiple dimensions agree on, flags key conflicts, and explains why they occur (e.g., timing, statistical definitions, sample differences, stakeholder stances, or unresolved facts).
  • Uncertainty and information gaps: Preserves uncertainty explicitly, specifies what cannot be answered robustly, and records what could overturn conclusions.
  • Final-deck wiring guidance: Provides structure and chapter allocation suggestions for the final report without drafting the reader-facing report itself.

Quick Start

Use the sn-research-synthesis skill to generate {report_dir}/synthesis.md from your existing {report_dir}/request.md, {report_dir}/plan.json, and all {report_dir}/sub_reports/*.md files.

Frequently Asked Questions about sn-research-synthesis

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I synthesize multiple research sub-reports into a single decision-ready judgment layer?

Research synthesis converts completed sub-reports into a structured judgment layer by extracting conclusion-relevant findings and producing 2-5 main-line judgments with evidence strength. It reads existing request and plan goals to generate a unified synthesis.md file.

What is the best way to resolve conflicts and track uncertainty across cross-dimensional research findings?

Cross-dimensional analysis resolves conflicts by summarizing consensus, flagging key disagreements, and explaining why they occur due to timing, sample differences, or stakeholder stances. Uncertainty tracking preserves gaps explicitly and records conditions that could overturn conclusions.

How do I label evidence strength and confidence levels for main-line research judgments?

Evidence grading assigns high, medium, or low confidence to each main judgment with justification and applicability conditions. This labeling ensures decision-makers understand the robustness of each finding extracted from the sub-reports.

What files do I need to generate a research synthesis markdown document?

Generating a synthesis.md requires reading request.md, plan.json, and all completed sub_reports/*.md files located in your report directory. These inputs provide the goals and scattered findings that the synthesis process converts into actionable judgments.

Can I use research synthesis to structure the final report deck without drafting the reader-facing content?

Research synthesis provides final-deck wiring guidance by offering structure and chapter allocation suggestions for the final report. It does not draft the reader-facing report itself, focusing instead on wiring the judgment layer to support downstream writing.

When should I avoid using automated research synthesis for deep research workflows?

Research synthesis is not suited for workflows lacking multiple completed sub-reports covering different dimensions. It requires existing sub-findings to extract conclusion-relevant threads and cannot generate judgments without prior dimensional research outputs.