review-loop

Iteratively review, validate, and repair code until findings are resolved.

1|Updated Mar 20, 2026
One-click install
npx skills add https://github.com/hirokimry/vibecorp --skill review-loop-hirokimry
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review-loop
Source: https://github.com/hirokimry/vibecorp/tree/main/skills/review-loop
Command: npx skills add https://github.com/hirokimry/vibecorp --skill review-loop-hirokimry

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill removes the back-and-forth of code review by repeatedly checking changes, validating findings, and applying fixes until the review is clean or the retry limit is reached.

Core Features & Use Cases

  • Review-to-fix automation: Runs an iterative loop that reviews changes, checks them against policy, plans repairs, and applies corrections.
  • Consensus handling for sensitive diffs: When full preset and trigger keywords match, it adds parallel analyst reviews and C-level meta-review for security, legal, and cost-related changes.
  • Safety and workflow control: Uses intent-aware severity triage, a maximum of five iterations, worktree-aware execution, and a completion stamp for PR readiness.

Quick Start

Ask the assistant to run the review-loop skill on the current changes and keep fixing issues until no findings remain.

Frequently Asked Questions about review-loop

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

FAQPage Schema
How do I automate fixing code review findings before a pull request is merged?

Automating code review fixes is handled by an iterative loop that validates findings, plans repairs, and applies corrections until changes are clean or the retry limit is reached. This removes manual back-and-forth during pull request cleanup.

What is consensus review for security and legal diffs?

Consensus review for sensitive diffs adds parallel analyst reviews and a meta-review layer when security, legal, or cost-related trigger keywords match. This ensures high-risk changes receive multi-layered validation before approval.

Can I run iterative code fixes in a git worktree without affecting my main branch?

Worktree-aware execution allows the iterative review and fix loop to run safely in a worktree without affecting your main branch. Changes are validated and repaired within the isolated worktree context.

How many retry iterations are applied during an automated pull request cleanup?

Automated pull request cleanup applies a maximum of five retry iterations to resolve review findings. The loop stops early if no findings remain, generating a review-ok stamp for PR readiness.

How does severity triage work during automated code review?

Severity triage uses intent-aware assessment to categorize review findings based on their impact and context. This prioritizes which issues the automated repair loop addresses first during validation.

What's the best way to handle repetitive code review fix loops for pull requests?

The best way to handle repetitive fix loops is coordinating iterative review, validation, and repair until findings are resolved or the bounded retry limit is reached. This workflow applies intent lookup, severity triage, and consensus aggregation.