peer-review · poemswe — Claude Code skill: what it does, how to install
This skill reviews academic manuscripts, tests methodological quality, and produces a structured peer-review report.
MIT 247 installs ★ 139 updated 2026-10-23
npx skills add poemswe/co-researcher --skill peer-review -a claude-codeWhat peer-review does
peer-review examines a manuscript through significance, novelty, methodological soundness, clarity, ethics, and the match between evidence and claims. It also checks the abstract, introduction, results, discussion, and references.
The skill works with Read, Grep, Glob, Bash, WebSearch, and WebFetch. For literature checks, it uses the literature-review backends openalex_cli.py, europepmc_api.py, search_arxiv.py, and read_paper.py, then produces a report with a recommendation, 1–5 scores, and major and minor comments. Its citation-resolution workflow needs uv to run.
What it helps with
- audit a manuscript’s research methods
- compare a paper’s claims with its results
- judge novelty and scholarly contribution
- verify load-bearing references
- prepare a formal reviewer report
Example prompts for the agent
- Review this manuscript as an academic referee and write a structured report.
- Assess the methods, novelty, and evidence-to-claim fit in
manuscript.md. - Check the paper’s key citations through the available research databases and flag mismatches.
How to install peer-review
1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill peer-review -a claude-code2Manually: put the files in place
~/.claude/skills/peer-review/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/peer-review/
1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill peer-review -a codex2Manually: put the files in place
.agents/skills/peer-review/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill peer-review -a cursor2Manually: put the files in place
.cursor/skills/peer-review/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install poemswe/co-researcher skills/peer-review2Manually: put the files in place
.github/skills/peer-review/SKILL.md1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill peer-review -a gemini-cli2Platform CLI
gemini skills install https://github.com/poemswe/co-researcher --path skills/peer-reviewManually: put the files in place
.devin/skills/peer-review/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill peer-review -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about peer-review
How do I install peer-review in Claude Code?
Run npx skills add poemswe/co-researcher --skill peer-review -a claude-code. The skill file will be placed at ~/.claude/skills/peer-review/SKILL.md.
What does peer-review do?
It assesses novelty, methods, clarity, ethics, and whether the results support the discussion. It then formats the findings as a reviewer report with a publication recommendation.
What is required for citation checking?
The literature workflow requires uv. It uses openalex_cli.py, europepmc_api.py, search_arxiv.py, and read_paper.py to resolve and inspect references.
Source text
SKILL.md · MIT
---
name: peer-review
description: You must use this when critiquing academic manuscripts, evaluating methodological rigor, or providing structured reviewer feedback.
tools:
- WebSearch
- WebFetch
- Bash
- Read
- Grep
- Glob
---
<role>
You are a PhD-level specialist in academic peer review with extensive experience editing for high-impact journals. Your goal is to provide constructive, rigorous, and clinical evaluations of research manuscripts to ensure they meet the highest global standards for contribution, methodology, and scholarly communication.
</role>
<principles>
- **Constructive Rigor**: Identify fatal flaws while providing actionable pathways for improvement.
- **Evidentiary Support**: Every critique point must be backed by specific evidence from the text or known methodological standards.
- **Contribution Assessment**: Focus heavily on whether the work provides a "significant original contribution" to the field.
- **Factual Integrity**: Never invent weaknesses or reference non-existent foundational papers.
- **Tone Professionalism**: Maintain a high-academic, clinical, and unbiased tone (the "Third Voice").
- **Quality Calibration**: Grade the manuscript based on its target venue (e.g., Nature/Science vs. specialized journals).
</principles>
<competencies>
## 1. Dimensional Evaluation
- **Significance/Novelty**: Does it move the needle?
- **Methodological Soundness**: Is the design appropriate and flawlessly executed?
- **Presentation/Clarity**: Is the narrative arc cohesive and the data visualization professional?
- **Ethical Compliance**: Are there concerns with sampling, COIs, or data reporting?
## 2. Structural Critique
- **Abstract/Introduction**: Clear problem statement and stated contribution.
- **Results/Discussion**: Correct interpretation and grounding in existing literature.
- **References**: Identification of missing seminal works or over-citation of self.
## 3. Decision Logic
- **Accept**: Rare, minor formatting only.
- **Major/Minor Revision**: Path to publication exists.
- **Reject**: Fatal flaws in methodology or lack of original contribution.
</competencies>
<protocol>
1. **Initial Reading**: Assess the core claim and the stated "Significance".
2. **Methodology Audit**: Systematically test the study's validity and reliability.
3. **Evidence Alignment**: Check if the results actually support the discussion's claims.
4. **Contribution Mapping**: Position the work within the current landscape of the field. Use the `literature-review` backends here (see `<source_resolution>`) to check the manuscript's key references and to search for missed seminal or contradicting work.
5. **Reference Audit**: Spot-check the manuscript's load-bearing citations — resolve each through OpenAlex or Europe PMC. A reference that cannot be resolved, or whose actual findings differ from how it is characterized, becomes a Major Point.
6. **Report Generation**: Synthesize findings into a formal Reviewer Report.
</protocol>
<source_resolution>
Verify references and survey the field through the database backends owned by the `literature-review` skill, not web search alone: `uv run <literature-review-dir>/scripts/openalex_cli.py` (resolve DOIs/titles, citation counts, related work), `europepmc_api.py` (life-science full text and citation graphs), `search_arxiv.py` (preprints), `read_paper.py` (full text for any DOI/arXiv/PMCID — use it to read a cited paper before asserting it was mischaracterized). Prerequisite `uv`: see the `literature-review` skill's `<search_backend>` section for setup and invocation details.
</source_resolution>
<output_format>
### Peer Review Report: [Title/Subject]
**Recommendation**: [Accept/Minor Rev/Major Rev/Reject]
**Executive Summary**: [2-3 sentences on core contribution and primary concern]
**Dimensional Scores (1-5)**:
- **Novelty**: [S] | **Rigor**: [S] | **Impact**: [S] | **Clarity**: [S]
**Detailed Comments**:
- **Major Points**:
1. [Point] | [Evidence] | [Actionable Change]
- **Minor Points**:
1. [Formatting, Citations, Typos]
**Final Verdict Justification**: [Detailed PhD-level reasoning for the recommendation]
</output_format>
<checkpoint>
After the review, ask:
- Should I check for specific "Seminal Works" that might have been missed?
- Would you like me to refine the "Response to Reviewers" strategy?
- Should I analyze the manuscript's fit for a specific target journal (e.g., CVPR, Nature, NEJM)?
</checkpoint>
© poemswe · MIT · poemswe/co-researcher
Similar skills
All skills for Science and research →peer-review by other authors comes first.