research-review

Automate multi-round external research critique via Codex MCP.

Updated Apr 29, 2026
One-click install
npx skills add https://github.com/jkfee/Auto-Research --skill research-review-jkfee
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: research-review
Source: https://github.com/jkfee/Auto-Research/tree/main/skills/research-review
Command: npx skills add https://github.com/jkfee/Auto-Research --skill research-review-jkfee

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides an automated, rigorous external critique of your research work by coordinating multi-round reviews via Codex MCP, surfacing weak claims, missing experiments, and narrative gaps.

Core Features & Use Cases

  • Multi-round critique: Orchestrates iterative feedback from an external reviewer to improve scientific rigor.
  • Weakness detection: Identifies logical gaps, unsupported claims, and missing experiments.
  • Narrative refinement: Helps structure a stronger story and clearer contribution for venues like NeurIPS/ICML.
  • Output generation: Produces a final consensus review document with actionable recommendations.
  • Use Case: Researchers preparing a manuscript or a project proposal can run a structured external review to strengthen claims and design.

Quick Start

Provide your project context and specific questions to initiate Round 1 with the Codex MCP reviewer.

Frequently Asked Questions about research-review

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

FAQPage Schema
How do I get automated external review for my research paper?

To get automated external review for your research paper, provide your project context, briefing, and specific questions to initiate a multi-round critique workflow via Codex MCP. The reviewer identifies logical gaps and unsupported claims, outputting a final consensus document.

What is multi-round critique for experimental design and how does it work?

Multi-round critique for experimental design works by orchestrating iterative feedback from an external reviewer backend via Codex MCP. It surfaces missing experiments and weaknesses across rounds, culminating in a final consensus review document with actionable recommendations.

Can I use AI-guided review to improve narrative analysis for conference submissions?

Yes, you can use AI-guided review to improve narrative analysis for conference submissions. The process structures a stronger story and clearer contribution for venues like NeurIPS/ICML by identifying narrative gaps and refining the overall manuscript presentation.

Do I need a Codex MCP server setup to run automated paper reviews?

Yes, you need a Codex MCP server setup to run automated paper reviews. The workflow requires configuring reviewer backends, feeding project context and questions to the server, and coordinating multi-round critiques to generate the final consensus feedback.

What is the best way to detect unsupported claims in a research report?

The best way to detect unsupported claims in a research report is by running a structured external review through Codex MCP. The multi-round critique workflow specifically targets weak claims, logical gaps, and missing experiments to strengthen scientific rigor.

When should I not use automated research review for my project?

You should not use automated research review when you lack a project briefing or cannot feed context and specific questions to the reviewer. The multi-round critique requires a Codex MCP server setup and structured project input to generate actionable feedback.