research-review

Automate iterative review of security, ZK, and LLM-security research papers against venue criteria.

Updated Aug 23, 2026

Search Tags:#LLM security #adversarial review #conference submission #research review #ZK proofs #security papers

Ready to run

npx skills add https://github.com/Jamie-Cui/opt --skill research-review-jamie-cui

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: research-review
Source: https://github.com/Jamie-Cui/opt/tree/main/dotfiles/skills/research-review
Command: npx skills add https://github.com/Jamie-Cui/opt --skill research-review-jamie-cui

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill automates the iterative review and improvement of security, ZK, and LLM-security research paper drafts, ensuring they meet the rigorous standards of top-tier security conferences.

Core Features & Use Cases

Adversarial Cross-Model Review: Simulates a skeptical senior PC member to identify overlooked flaws.
Security Venue Criteria Alignment: Targets specific criteria for conferences like CCS, USENIX, and S&P.
Iterative Improvement Loop: Replaces extensive experiments with proof sketches, threat model analysis, and literature gap fixing.
Use Case: A researcher can use this skill to refine a novel ZK proof system paper, getting automated feedback on its threat model, security claims, and novelty before submission.

Quick Start

Use the research-review skill to perform 4 rounds of automated review on the paper draft located in the current directory.

Frequently Asked Questions about research-review

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

FAQPage Schema

How do I automate iterative review for security research papers before conference submission?▼

Security research papers are refined for top-tier venue submission through automated iterative review. This process executes adversarial cross-model loops to identify flaws, replacing extensive GPU experiments with proof sketches and threat model analysis.

What is adversarial cross-model review for LLM security papers?▼

Adversarial cross-model review for LLM security papers simulates a skeptical senior PC member to identify overlooked flaws. It requires dispatching sub-agents or alternative LLM backends to rigorously challenge security claims and threat models.

Does this iterative paper review align with specific security conference criteria like CCS or USENIX?▼

Yes, iterative paper review aligns with specific security conference criteria for venues like CCS, USENIX, and S&P. It targets these rigorous standards by evaluating novelty, threat models, and security claims against exact submission requirements.

How do I refine ZK proofs in a research paper without running extensive GPU experiments?▼

Refine ZK proofs without GPU experiments by replacing them with proof sketches and threat model analysis. The iterative review loop evaluates zero-knowledge security claims and literature gaps to ensure mathematical arguments meet top-tier venue standards.

Do I need alternative LLM backends to perform adversarial review on security papers?▼

Alternative LLM backends are required to perform adversarial review on security papers. The skill depends on dispatching sub-agents to these different backends to execute cross-model review loops and effectively simulate skeptical PC members.

What are the limitations of using automated review loops for LLM security research?▼

Limitations of automated review loops for LLM security research include the strict dependency on sub-agent dispatch for adversarial reviews. It replaces empirical GPU experiments with proof sketches, which may not fully validate complex experimental claims without human verification.