From academic-writing
Simulate peer review of a paper. Use when you want to anticipate reviewer concerns, identify weaknesses, or get a simulated review with scores before submission. Trigger whenever the user asks to review a paper, check for weaknesses, simulate what reviewers would say, do a pre-submission check, stress-test a draft, or get feedback on a manuscript — even informally like "what would reviewers think of this" or "is this ready to submit."
How this skill is triggered — by the user, by Claude, or both
Slash command
/academic-writing:peer-reviewThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Produce a structured simulated review that anticipates real reviewer concerns, identifies weaknesses, and provides actionable improvement suggestions.
Produce a structured simulated review that anticipates real reviewer concerns, identifies weaknesses, and provides actionable improvement suggestions.
Before reviewing, read references/reviewer-guidelines.md for reviewer evaluation criteria and guidelines.
Choose the appropriate review mode based on what the user needs:
Goal: Help strengthen the paper before submission. Balanced, constructive, identifies fixable weaknesses.
Use when the user says things like "review my paper," "is this ready to submit," "what should I improve."
Goal: Simulate the harshest plausible reviewer. Find every weakness, question every assumption, demand more evidence.
Use when the user explicitly wants to stress-test (e.g., "what's the worst a reviewer could say," "tear this apart," "find every weakness").
The review format is the same for both modes — the difference is in how aggressively you probe for weaknesses and how high you set the bar.
Read the paper end-to-end before forming judgments. Note first impressions but reserve scoring until after careful analysis.
| Dimension | What to Evaluate |
|---|---|
| Quality | Technical soundness, well-supported claims, correct methodology, reproducibility |
| Clarity | Clear writing, logical flow, reproducible by experts in the field |
| Significance | Community impact, advances understanding, practical or theoretical importance |
| Originality | New insights, methods, or perspectives (doesn't require entirely new method) |
Systematically check for:
| Weakness Type | What to Look For |
|---|---|
| Overclaiming | Claims not supported by evidence; gap between abstract/intro claims and actual results |
| Weak baselines | Missing obvious comparisons; outdated baselines; unfair comparison setup |
| Unclear contributions | Can't identify what's novel; contribution buried or vague |
| Missing ablations | No analysis of which components matter; can't tell what drives performance |
| Reproducibility concerns | Missing hyperparameters, code, dataset details, or compute information |
| Writing clarity | Hard to follow; inconsistent terminology; missing signposting |
| Insufficient evaluation | Too few datasets; no error bars; no statistical significance testing |
| Missing limitations | No discussion of when the method fails or its scope |
Even in pre-submission mode, consider what a skeptical reviewer would challenge:
Simulated reviews are only useful if they're calibrated to real reviewer behavior. Common pitfalls:
| Pitfall | Why It's Bad | Correction |
|---|---|---|
| Too generous | Authors submit overconfident, get blindsided | Be honest about weaknesses — that's the point of simulation |
| Too harsh | Authors lose confidence in good work | Distinguish major weaknesses (would change recommendation) from minor issues (should fix but not deal-breakers) |
| Focusing on minor issues | Misses the forest for the trees | Lead with the 2-3 things that would most influence a real reviewer's score |
| Ignoring the positive | Real reviews acknowledge strengths | Always identify genuine strengths — this helps authors know what's working |
Rule of thumb: If you can't identify at least 2 genuine strengths, either the paper has fundamental problems or you haven't read it carefully enough.
## Summary of Contributions
[1-2 sentences: what the paper claims to contribute, as the reviewer understands it]
## Strengths
- [Strength 1]
- [Strength 2]
- [Strength 3]
(2-4 bullet points, specific and evidence-based)
## Weaknesses
- [Weakness 1 — severity: major/minor]
- [Weakness 2 — severity: major/minor]
- [Weakness 3 — severity: major/minor]
(2-4 bullet points, ranked by severity, with specific suggestions for how to address each)
## Questions for Authors
- [Question 1]
- [Question 2]
(Questions whose answers could change the assessment)
## Overall Assessment
Score: [X/10]
Confidence: [X/5] (1 = not confident, 5 = very confident)
Justification: [2-3 sentences explaining the score]
## Actionable Suggestions
- [Specific improvement 1]
- [Specific improvement 2]
(Concrete steps to address weaknesses before submission)
Use a 1-10 scale unless the user specifies a different scale:
| Score | Label | Meaning |
|---|---|---|
| 9-10 | Strong Accept | Exceptional work; among the best you'd expect at a top venue |
| 7-8 | Accept | Solid contribution with minor issues; would strengthen the venue |
| 5-6 | Borderline | Has merit but significant weaknesses; could go either way |
| 3-4 | Below threshold | Fundamental issues with claims, methodology, or evaluation |
| 1-2 | Strong Reject | Critical flaws, known results, or ethical concerns |
Confidence scale:
| Score | Meaning |
|---|---|
| 5 | Very confident — deep expertise in this exact area |
| 4 | Confident — familiar with the field and related work |
| 3 | Somewhat confident — know the area but not the specific subfield |
| 2 | Low confidence — outside main expertise |
| 1 | Guessing — very unfamiliar with the topic |
Use this as a quick scan before writing the review:
references/reviewer-guidelines.md for evaluation guidelines and common concernsnpx claudepluginhub jasonbian97/jason-cc-skills --plugin academic-writingGenerates structured peer review reports for academic manuscripts, evaluating novelty, methodological rigor, clarity, impact, and ethics. Use when critiquing papers or providing reviewer feedback.
Simulates a full academic journal peer review with EIC decision and 3 independent reviewers (Methodology Expert, Domain Specialist, Devil's Advocate). Scores 1–10 and issues verdicts: Accept, Minor Revision, Major Revision, or Reject.
Simulates 5 independent academic peer reviewers (Editor-in-Chief, 3 peers, Devil's Advocate) for multi-perspective paper review. Supports full review, re-review, quick assessment, methodology-focus, and Socratic guided modes.