skills.rest
OpenLAIR
OpenLAIROFFICIAL
Explore all skills by this author →
GitHub

aris-auto-review-loop

Automates iterative research review cycles using Codex MCP until positive assessment.

1.1k|116|Updated Feb 26, 2026
Search Tags:#Automation#iterative improvement#peer review#state persistence#research review#codex mcp#Machine learning
One-click install
npx skills add https://github.com/OpenLAIR/dr-claw --skill aris-auto-review-loop
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: aris-auto-review-loop
Source: https://github.com/OpenLAIR/dr-claw/tree/main/skills/aris-auto-review-loop
Command: npx skills add https://github.com/OpenLAIR/dr-claw --skill aris-auto-review-loop

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Manually iterating on research work through repeated external reviews is slow and inconsistent. This Skill automates the full review-fix-re-review loop, driving an external GPT reviewer via Codex MCP until the work reaches a positive assessment or a maximum round limit.

Core Features & Use Cases

  • Autonomous Review Loop: Repeatedly sends research context to an external reviewer, parses scores and weaknesses, implements fixes, and re-reviews up to 4 rounds.
  • Adjustable Reviewer Difficulty: Three modes (medium, hard, nightmare) ranging from curated MCP review to adversarial review where GPT reads the repository directly and verifies claims against code.
  • State Persistence and Recovery: Writes REVIEW_STATE.json after each round so long-running loops survive context compaction and can resume mid-run.
  • Use Case: A researcher finishing an ML paper runs the loop overnight; each round the reviewer scores the work, weaknesses are fixed (new experiments, analyses, reframing), and AUTO_REVIEW.md accumulates a full audit trail ending with claims ready for paper writing.

Quick Start

Ask the AI to run an auto review loop on your current research project until the reviewer accepts it.

Frequently Asked Questions about aris-auto-review-loop

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

FAQPage Schema
How do I run an autonomous review loop for my research project?▼

Invoke the skill with a topic or scope, and it will send your research context to an external GPT reviewer via Codex MCP, parse the score and weaknesses, implement fixes, and re-review until the verdict is positive or 4 rounds complete.

What is the difference between medium, hard, and nightmare reviewer difficulty?▼

Medium uses standard MCP review where Claude curates context. Hard adds persistent reviewer memory and a debate protocol where Claude can rebut criticisms. Nightmare lets GPT read the repository directly via codex exec to independently verify claims against code and results.

Can the review loop recover after context window compaction?▼

Yes. The loop writes REVIEW_STATE.json after every round with the round number, thread ID, score, and pending experiments. On restart, it reads this file and AUTO_REVIEW.md to resume from the next round if the state is less than 24 hours old.

When does the auto review loop stop iterating?▼

The loop stops when the reviewer gives a score of 6 or higher with a verdict of ready or almost, or when the maximum of 4 rounds is reached. Users can also stop it early via the optional human checkpoint mode.

Does the skill fabricate citations when fixing reviewer concerns?▼

No. It enforces an anti-hallucination citation chain: BibTeX entries are fetched from DBLP first, then CrossRef as fallback, and unresolvable references are marked with a VERIFY flag rather than generated from memory.

Related Skills

Explore More Education & Research Skills→
KYRIE66nbKYRIE66nb

paper-plan

Generate structured paper outlines from review conclusions and experiment results.

Community
Advanced
RheasilviaRheasilvia

research-paper-writing

Automate machine learning research from experiment design to LaTeX drafting.

Community
Advanced
ChuaHanChongChuaHanChong

monitor

Poll ML training log files to detect divergence and terminate failing processes.

Community
Advanced
jandan138jandan138

auto-review-loop-minimax

Automate iterative ML research peer review with MiniMax across MAX_ROUNDS.

Community
Advanced