paperjury

Reviews and revises CS-conference papers through an adversarial multi-agent courtroom engine.

1.2k|43|Updated Jun 2, 2026
One-click install
npx skills add https://github.com/Spark-To-Paper-Skills/paperjury --skill paperjury-spark-to-paper-skills
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: paperjury
Source: https://github.com/Spark-To-Paper-Skills/paperjury
Command: npx skills add https://github.com/Spark-To-Paper-Skills/paperjury --skill paperjury-spark-to-paper-skills

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Before submitting a CS-conference paper, authors need to know which reviewer-style criticisms are real, which are misreadings, and which require new experiments. PaperJury runs a closed review-verdict-revise-verify loop on LaTeX, Markdown, or Word manuscripts so every issue is adjudicated instead of blindly applied, and safe fixes land only behind author sign-off. ## Core Features & Use Cases - Adversarial courtroom review: N holistic domain reviewers read the whole paper, contested issues go to a two-sided jury trial, and each charge terminates as valid-fixable, author-required, or invalid-drop, tracked in a durable JSON ledger. - Guarded revision with real verification: minimal patches are screened by deterministic guards (anchor-diff, cross-reference checks, real LaTeX compile, desk-reject compliance screening) and risky edits queue for the author instead of being applied silently. - Three modes: direct-edit for single asked-for changes, review for pre-submission hardening, and auto for unattended multi-round loops under an explicit /goal. - Use Case: You have a 21-page NeurIPS draft. Ask for a full review; PaperJury dedupes 152 reviewer complaints into 55 issues, safely applies 26 fixes, hands 10 back to you, drops 19 as invalid, and reports a clean compile. ## Quick Start Ask the assistant to review this paper with paperjury, focusing on whether the experiments support the claims.

Frequently Asked Questions about paperjury

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

FAQPage Schema
How do I get an AI review of my paper before submission?▼

Ask for a review or mock-review of your manuscript and PaperJury runs its courtroom engine: domain reviewers read the whole paper, contested issues go to a jury trial, and each issue ends as valid-fixable, author-required, or invalid-drop in a tracked ledger.

Can PaperJury review a Word docx file?▼

Yes, PaperJury converts a .docx once into a Markdown working copy with an honesty report of what was kept or dropped, then runs the full multi-round review on that copy. The original Word file is never modified.

Does PaperJury edit my paper automatically?▼

No. In direct-edit and review modes every patch requires your explicit sign-off before it is applied. Auto mode must be enabled explicitly via /goal and still queues risky edits for one human pass.

What tools does PaperJury need installed?▼

Node.js is required because all deterministic guards (ledger, apply-patch, compile-guard, compliance-check) run on it. A LaTeX toolchain is optional; without it the compile check degrades to a structural lint reported as compiled:null.

When should I not use PaperJury?▼

Do not use it to write a paper from scratch, generate figures, or produce an official-venue rebuttal. It is a pre-submission self-check loop and cannot invent experiments, fabricate results, or replace peer review.