# Synthesize Evidence Brief

> Synthesize multi-source evidence into one bounded brief with residuals, conflicts, and the next observation worth buying. Use when a consequential question needs a traceable literature or evidence synthesis. Do not use for a single-claim critique, a domain go/no-go review, or ordinary product implementation.

- Skill: `sylphxai/synthesize-evidence-brief` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add sylphxai/synthesize-evidence-brief`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sylphxai/synthesize-evidence-brief/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: SylphxAI (https://skillmd.com/u/sylphxai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/sylphxai/synthesize-evidence-brief

---


# Synthesize Evidence Brief

The review procedure stays smaller than the decision it protects. Define the question, the decision it would change, and what evidence would reverse the conclusion before searching.

Extract comparable design, population, exposure, outcome, estimate, uncertainty, and conflicts. Source dependence, missingness, and selective reporting move certainty before extra citations. Pool only when the measures and contexts support a combined meaning. Residual unknowns stay unknown.

Open [reproducible evidence synthesis](references/reproducible-evidence-synthesis.md) when search depth, bias appraisal, or pooling rules need depth.

Use `analyze-critically` for one disputed claim. Use `review-domain` for a named product go/no-go. Use `analyze-causal-inference` when the brief must identify a causal effect rather than summarize reports.