stage-peer-reviewer

Simulates a five-perspective peer review panel that scores manuscripts against anchored rubrics.

2|Updated Aug 3, 2026
One-click install
npx skills add https://github.com/wanghao9610/STAGE --skill stage-peer-reviewer-wanghao9610
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: stage-peer-reviewer
Source: https://github.com/wanghao9610/STAGE/tree/main/.qwen/skills/stage-peer-reviewer
Command: npx skills add https://github.com/wanghao9610/STAGE --skill stage-peer-reviewer-wanghao9610

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Researchers submitting papers lack early, honest feedback on how a real program committee would judge their manuscript, and informal self-review misses weaknesses in novelty, soundness, rigor, and clarity. ## Core Features & Use Cases - Five-Perspective Panel: Dispatches concurrent reviewers covering novelty and related work, technical soundness, experimental rigor and reproducibility, clarity and presentation, plus a devil's advocate, synthesized by a chair into one meta-review. - Anchored Rubric Scoring: Scores on a 6-point conference scale or journal tiers with hard caps and honest confidence levels, never averaging panel enthusiasm. - Citation Integrity Contract: References must come from the manuscript's whitelist or be verified by logged searches during the run; unverifiable claims are demoted to unnamed directions. - Use Case: Before submitting to a conference, run the panel on your built PDF to receive a SIM_REVIEW report whose weaknesses name the exact claim IDs they attack, then route fixes to drafting and auditing skills or draft a rebuttal with the response writer. ## Quick Start Ask the assistant to run a simulated peer review panel on this repository's manuscript for the current cycle, optionally specifying quick mode or an external paper path.

Frequently Asked Questions about stage-peer-reviewer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a simulated peer review on my paper before submission?

Invoke the skill with the default panel mode against your repository's manuscript. It builds the PDF first, dispatches five perspective reviewers concurrently, and writes a meta-review to the cycle's reviews directory shaped like a real review.

What is the difference between panel mode and quick mode?

Panel mode dispatches five independent concurrent reviewer delegates, one per perspective, while quick mode has the chair walk all five perspectives sequentially in a single pass. Quick mode is cheaper but the review is marked as not independent.

Can I use this to referee someone else's paper?

Yes, pass extern=<path> pointing at the external PDF or source file. The same panel and rubric apply, but there is no build step, no claim ledger, confidential search is on by default, and the report is written under wkdrs/ instead of the reviews directory.

Does the simulated review edit my manuscript or claim ledger?

No. The skill never edits the manuscript, the claim ledger, or venue configuration. Its only durable output is the meta-review file, and the manuscript leaves the run byte-identical to how it entered.

What happens if my paper fails to build before review?

A failed build becomes finding number one in the review. The panel then proceeds over the manuscript sources, with the partiality of reviewing without a built PDF stated explicitly in the Summary and Synthesis Notes.

How are references handled during the simulated review?

Every named reference must be in the manuscript's bibliography whitelist or verified by a logged search during the run. The chair re-verifies each claimed hit against the criterion written before the query, and unverifiable references are demoted to unnamed directions.