skills.rest
Shallow-W
Shallow-WCOMMUNITY
Explore all skills by this author →
GitHub

auto-review-loop

Automate iterative research review cycles with score-driven stopping conditions.

Updated Apr 21, 2026
Search Tags:#quality assurance#acceptance criteria#research review#autonomous improvement#paper iteration#codex mcp#experiment fixes
One-click install
npx skills add https://github.com/Shallow-W/llm-wiki --skill auto-review-loop-shallow-w
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: auto-review-loop
Source: https://github.com/Shallow-W/llm-wiki/tree/main/.claude/skills/auto-review-loop
Command: npx skills add https://github.com/Shallow-W/llm-wiki --skill auto-review-loop-shallow-w

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill solves the problem of getting research work accepted by running an autonomous multi-round review loop that repeatedly critiques, fixes, and re-validates until the work is ready.

Core Features & Use Cases

  • Autonomous review loop: Repeatedly sends your research state to an external reviewer, parses the score/verdict, then applies the minimum required fixes and re-runs the cycle.
  • Configurable reviewer behavior: Supports medium (MCP review), hard (adds reviewer memory + debate), and nightmare (adversarial reviewer with direct repo reads via codex exec).
  • Persistent state & auditability: Saves review logs, maintains compaction recovery state in REVIEW_STATE.json, and versions outputs via the shared protocols to preserve traceability.

Quick Start

Run the auto-review-loop skill for your research topic when you want the system to review, implement fixes, and re-review until it reaches an accepted/ready verdict.

Frequently Asked Questions about auto-review-loop

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

FAQPage Schema
How do I automate iterative paper review and apply experiment fixes until acceptance criteria are met?▼

Automated iterative paper review works by repeatedly sending research state to an external reviewer, parsing the score and verdict, applying minimum required fixes, and re-reviewing until a readiness verdict is reached.

What is an autonomous review loop for research preparation and how does it handle weaknesses?▼

An autonomous review loop for research preparation repeatedly critiques claims, identifies weaknesses, applies minimum fixes, and re-validates the work across multiple rounds with score-driven stopping conditions until accepted.

How do I configure adversarial quality checks with reviewer memory and debate for research review?▼

Adversarial quality checks can be configured using hard or nightmare modes, adding reviewer memory and debate, or enabling an adversarial reviewer with direct repository reads via codex exec to apply stricter validation.

Can I use Codex MCP for reviewer routing in a multi-round research review process?▼

Yes, Codex MCP is used for reviewer routing within the multi-round research review process, sending your research state to an external reviewer and parsing the structured score, verdict, and action items returned.

How do I recover round-state if an automated research improvement loop fails midway?▼

Round-state recovery is handled by maintaining compaction recovery state in REVIEW_STATE.json, preserving persistent state and auditability so the automated research improvement loop can resume without losing progress.

What are the limitations of using YAML-configured limits for autonomous paper iteration?▼

YAML-configured limits restrict the autonomous paper iteration process by defining boundaries for the multi-round review loop, meaning the system will stop applying experiment fixes once these configured score-driven stopping conditions are met.

Related Skills

Explore More Education & Research Skills→
raja21068raja21068
2

auto-review-loop

Automate iterative research-code review and repair until acceptance criteria are met.

Community
Advanced
goupup-aigoupup-ai
2

auto-review-loop

Automate iterative research review and improvement cycles with configurable reviewer backends.

Community
Advanced
zhuyingqinzhuyingqin
1

auto-review-loop

Run iterative research review loops with reviewer scores and verdicts.

Community
Advanced
THUFanZdTHUFanZd

auto-review-loop-llm

Automate multi-round research review and fix iterations via OpenAI-compatible LLM APIs.

Community
Advanced