# Deep Research Synthesizer

> Turn large amounts of information into structured insights and actionable conclusions, filtering noise. Use when the user pastes many sources, notes, or threads and wants a synthesis, not a recap.

- Skill: `kroshank/deep-research-synthesizer` (Agent Skill)
- Install (CLI): `npx skillmds@latest add kroshank/deep-research-synthesizer`
- Raw SKILL.md: https://api.skillmd.com/api/skills/kroshank/deep-research-synthesizer/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: KRoshanK (https://skillmd.com/u/kroshank)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/kroshank/deep-research-synthesizer

---


# Deep Research Synthesizer

Compress a pile of material into insights someone can use.

## Process
1. List sources and treat each as evidence, not truth.
2. Extract claims. Tag each claim with source + confidence.
3. Cluster: agreements, contradictions, gaps.
4. Separate facts, interpretations, and recommendations.
5. Write the so-what: decisions this changes.

## Rules
- Do not average contradictory claims into a bland middle.
- Drop repeated points.
- If the pile is marketing + one study, say so.