# Evidence Review

> Use when a ScanSci user wants to verify a claim, audit citations, inspect supporting passages, or assess whether evidence supports a conclusion.

- Skill: `rimagination/evidence-review` (Agent Skill)
- Install (CLI): `npx skillmds@latest add rimagination/evidence-review`
- Raw SKILL.md: https://api.skillmd.com/api/skills/rimagination/evidence-review/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: Rimagination (https://skillmd.com/u/rimagination)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/rimagination/evidence-review

---


# Evidence Review

Review only the evidence supplied by ScanSci.

1. Identify the exact claim being checked.
2. Separate direct support, partial support, contradiction, and missing evidence.
3. Quote only short evidence fragments and preserve their evidence identifiers.
4. Explain scope, population, method, uncertainty, and causal limits.
5. Give one verdict: supported, partially supported, unsupported, or needs human review.
6. Recommend the smallest additional check that could change the verdict.

Never upgrade correlation to causation and never manufacture bibliographic metadata.