skills.rest
Mang30
Mang30COMMUNITY
Explore all skills by this author →
GitHub

auto-review-loop

Iteratively review, fix, and re-review research work until readiness thresholds are met.

Updated May 29, 2026
Search Tags:#research#experimental validation#paper review#state recovery#autonomous improvement#iterative debugging#quality threshold
One-click install
npx skills add https://github.com/Mang30/myskills --skill auto-review-loop-mang30
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: auto-review-loop
Source: https://github.com/Mang30/myskills/tree/main/skills/auto-review-loop
Command: npx skills add https://github.com/Mang30/myskills --skill auto-review-loop-mang30

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves the problem of getting research work reliably reviewed by repeatedly running a full review-fix-review loop until the work meets a clear external “ready/almost ready” bar or a maximum number of rounds is reached.

Core Features & Use Cases

  • Autonomous multi-round review loop: Iterates review → implement fixes → re-review across multiple rounds with a hard stop at MAX_ROUNDS.
  • Structured scoring and decisioning: Enforces a POSITIVE_THRESHOLD based on score and verdict keywords to determine when to stop.
  • Optional robustness upgrades: Supports reviewer difficulty modes (medium/hard/nightmare), including reviewer memory and fully adversarial repo verification when configured.
  • Round documentation and state recovery: Appends complete reviewer raw responses to a cumulative log and persists state to REVIEW_STATE.json for compaction-safe resumption.

Quick Start

Use the auto-review-loop skill to review and improve a project by running /auto-review-loop "your topic or scope".

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 experimental validation until a readiness threshold is met?▼

Autonomous iterative research improvement works by repeatedly performing external review, applying minimum fixes, and re-running the review loop until a readiness threshold is met or maximum rounds are reached.

What is an autonomous review-fix loop for multi-round research manuscript refinement?▼

An autonomous review-fix loop iterates across multiple rounds, executing full reviews, implementing fixes, and re-reviewing with a hard stop at a defined maximum number of rounds.

How does state recovery work for iterative debugging when a review loop is interrupted?▼

State recovery for iterative debugging works by persisting progress to a REVIEW_STATE.json file, enabling compaction-safe resumption of the review loop from the last saved state.

Can I use adversarial verification modes for checking research claims across analysis and code?▼

Yes, adversarial verification of claims is supported through optional reviewer difficulty modes including hard and nightmare settings, featuring reviewer memory and fully adversarial repository verification.

What is the best way to enforce a structured scoring threshold for research paper readiness?▼

Structured scoring for research paper readiness is enforced by a positive threshold based on score values and verdict keywords to determine exactly when the iterative review loop should stop.

Do I need YAML configuration to define operational parameters for autonomous research improvement?▼

Yes, YAML-defined operational parameters are required to configure the autonomous iterative research improvement loop, including thresholds, maximum rounds, and optional reviewer difficulty modes.

Related Skills

Explore More Education & Research Skills→
hve4638hve4638

auto-review-loop

Automate multi-round research artifact review and improvement via Codex MCP.

Community
Advanced
shishishishi
3

codex-review

Automate iterative code reviews with Codex and Claude Code patches.

Community
Advanced
hoonsubinhoonsubin

doublecheck

Extract claims from AI-generated text and verify them via web search.

Community
Advanced
HeXiao-55HeXiao-55
1

auto-review-loop

Automate multi-round research reviews with iterative critique and fixes via Codex MCP.

Community
Advanced