domains / research / manuscript-peer-review / peer-review

peer-review

Skill card

combined

The copies overlap on the core task of structured scientific peer review, but they disagree on the procedure and contain material steps the others lack. The K-Dense-AI copy is the only one that enforces a mandatory safety boundary, authorization gate, and local-only processing before reading unpublished content, and it provides specific bundled CLI scripts for intake validation and reporting-guideline selection. The Vibe-Skills and brycewang-stanford copies are nearly identical to each other and add a 'Visual Enhancement' step requiring the generation of scientific schematics, which the K-Dense copy explicitly forbids (no image pipelines). The aipoch copy provides a distinct, simpler workflow focused on pre-submission checks and a specific 'Major vs. Minor' issue triage template that the others do not use. Crowning one copy would hide the safety-critical constraints of the first or the visual/template-specific steps of the others.

Repositories

Other copies

Combined skill

# Peer Review

## 1. Safety and Authorization (Mandatory)
Before reading or analyzing unpublished content:
1. Confirm the user is authorized by the publisher, editor, or author.
2. Check the target venue’s review, confidentiality, and AI/tool policies.
3. Record conflicts of interest and competence limits.
4. Default to local-only processing. Do not send unpublished content to external services or image generators without specific authorization.

## 2. Intake and Scope
1. Establish the submission type, stage, and target venue.
2. Identify available materials (manuscript, supplements, data/code).
3. Record missing material that prevents assessment.

## 3. Initial Assessment
1. Read the abstract and full text to form an overall impression.
2. Identify the research question, claimed contributions, and novelty.
3. Provide a brief summary (2-3 sentences) of the manuscript's essence.

## 4. Detailed Section-by-Section Review
Evaluate each section using these criteria:
- **Abstract/Title:** Accuracy, clarity, and completeness.
- **Introduction:** Context, rationale, novelty, and literature coverage.
- **Methods:** Reproducibility, rigor, ethics, and statistical appropriateness. Verify sample sizes, randomization, and controls.
- **Results:** Logical presentation, statistical reporting (effect sizes, CIs), and objectivity.
- **Discussion:** Interpretation, alignment with results, and acknowledgment of limitations.

## 5. Reporting Guidelines and Evidence Mapping
1. Select the appropriate reporting guideline (e.g., CONSORT, STROBE) for the study design.
2. Map central claims to supporting evidence (figures, tables, citations).
3. Check for alignment between claims and results.

## 6. Issue Organization and Triage
Classify findings into:
- **Major Issues:** Validity threats, methodological flaws, unsupported claims, or ethical gaps. For each, provide: Why it matters, Suggested fix, and Expected impact.
- **Minor Issues:** Clarity, formatting, citations, or language improvements. For each, provide: Suggested fix.

## 7. Recommendation
1. Provide a decision (Accept / Minor Revision / Major Revision / Reject).
2. Explain the rationale based on novelty, rigor, and evidence strength.
3. List the top 3-5 changes that would most improve the manuscript.

## 8. Visual Enhancement (Optional/Authorized)
If the venue policy and authorization permit, and the user requests it:
- Generate scientific schematics to visualize workflows, decision trees, or complex concepts.
- Ensure diagrams are publication-quality and accessible.