PaperJury
Evidence-linked pre-submission review and bounded revision for CS-conference manuscripts.
Last reviewed by the paperbanana team on Oct 20, 2026
Install
/plugin marketplace add Spark-To-Paper-Skills/paperjury
/plugin install paperjury@Spark-To-Paper-SkillsWhat is PaperJury?
PaperJury is a pre-submission editing and review skill for computer-science conference papers in vision, NLP, and ML. Direct-edit mode handles a requested passage change; review mode organizes a multi-perspective review into an evidence-linked issue ledger. Each concern can be judged invalid, valid and fixable, or requiring the author's research judgment.
The package includes deterministic Node scripts for issue state, patch handling, and checks around revision. Safe textual fixes are separated from issues requiring new data, experiments, or author knowledge. Auto mode is opt-in and uses an agreed revision policy; completing a review round can still leave author-required issues unresolved.
LaTeX and Markdown are native working formats. DOCX intake extracts a Markdown working copy and records conversion losses; it does not edit the original Word file. PDF is not a supported native input. The repository includes examples and reported evaluations, but those results were not independently reproduced for this directory review.
Core capabilities
Three editing modes
Use direct-edit for a passage, review for pre-submission critique, or explicitly opt into a bounded unattended revision loop.
Adjudicated review issues
Route concerns through review and, where needed, adversarial deliberation rather than accepting every generated criticism.
Persistent revision ledger
Track issue evidence, locations, verdicts, and lifecycle state across rounds while keeping author-required items visible.
Patch and verification guards
Prepare minimal changes and check the resulting manuscript; report unavailable LaTeX compilation rather than treating it as a pass.
What you can use it for
Stress-test a CS-conference draft
Ask reviewers to inspect whether the experiments support the paper's claims before submission.
Tighten a passage without changing its meaning
Use direct-edit mode and inspect the proposed patch before it is applied.
Separate edits from new research
Collect missing ablations, evidence gaps, and disputed decisions in an author queue while handling supported textual fixes.
How to get started
- 1
Install the Claude Code plugin
Run the marketplace and plugin commands above, then start a session in the manuscript project. The README links a separate Codex edition.
- 2
Choose the manuscript and scope
Provide LaTeX or Markdown and identify the target venue family. For DOCX, inspect the extraction report and Markdown working-copy path.
- 3
Start with review or direct-edit
Describe the desired review or passage change. Inspect the issue ledger or proposed patch and supply author decisions.
- 4
Verify the revised working copy
Check compilation when a TeX toolchain is available and read the outstanding author-required queue. Enable auto mode only with an explicit revision policy.
How it compares to similar skills
PaperJury is a CS-conference manuscript review-and-revision tool, not a from-scratch drafting system or an official peer-review service.
AI Research Feedback
Choose AI Research Feedback for economics or finance papers, pre-analysis plans, grant proposals, and empirical paper–code checks.
Academic Research Skills
Choose Academic Research Skills for a broader research-to-writing pipeline rather than a dedicated adjudication and revision loop.