Review Loop

Review-loop automates the draft-critique-revise cycle that most people do by hand when they're unhappy with a first pass. You kick it off with a phrase like "use review-loop" or "/review-loop", the main agent does the work, then spawns a fresh reviewer subagent that sees only the artifact, not the reasoning behind it, and scores it 1 to 10 with file-and-line feedback. It revises until the score clears a gate (8 by default) within a 2-to-4 loop window. The part worth noticing is the model rule: the reviewer has to be at least as capable as the worker, since a weaker critic can't catch subtle regressions. The skill also includes an escalation path if scores stall for two loops. It's a good fit for specs, security-sensitive code, and financial logic, and overkill for one-line fixes. The fresh-context reviewer is the design choice that makes it more than self-editing.

skills.sh installs: 110.9K
GitHub stars: 7
last updated: Oct 26, 2026

View on GitHub

Install

$ npx -y skills add 2dmurali/review-loop-skill --skill review-loop --agent claude-code

Installs into .claude/skills of the current project.

Files

Review Loop

Iterative worker-reviewer cycle within a single session. You do the work, spawn a reviewer subagent to critique it, revise based on feedback, repeat until quality gate is met.

Core principle: First drafts are never final. Iterative critique produces better output than a single pass.

Platform note: This skill works best with agents that support subagent spawning. On platforms without that capability, simulate the reviewer step by opening a fresh chat context, pasting only the work product (no prior reasoning), and asking it to score 1-10 with specific feedback.

Quick Start

Say: "implement X, use review-loop" or "run review-loop on the file I just wrote"
The agent does the work (or reads existing work)
A separate critic subagent scores it 1-10 with specific feedback
The agent revises and repeats until score >= 8 (default quality gate)
A loop summary is delivered with the final output