roborev-refine

Automate iterative fixing of code review findings in git feature branches.

Updated Feb 5, 2026
One-click install
npx skills add https://github.com/machow/dev-skills --skill roborev-refine
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: roborev-refine
Source: https://github.com/machow/dev-skills/tree/main/roborev-refine
Command: npx skills add https://github.com/machow/dev-skills --skill roborev-refine

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates the process of diagnosing and fixing code review findings by continuously iterating until all reviews pass, enabling hands-off code quality improvement.

Core Features & Use Cases

  • Finds failed reviews on the current branch
  • Runs an agent to fix findings
  • Commits the fixes
  • Waits for re-review
  • Repeats up to 10 iterations by default
  • Runs a branch-level review once per-commit fixes pass
  • Addresses branch-level findings as well
  • The agent runs in an isolated worktree to keep the working directory clean.

Quick Start

Run roborev refine to start the automated review-fix loop on your current feature branch.

Frequently Asked Questions about roborev-refine

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

FAQPage Schema
How do I automate fixing code review findings on a git branch?

Automating code review fixes involves running an iterative agent that diagnoses findings, applies fixes, commits them, and waits for re-review until all checks pass. This process targets feature branches and repeats up to 10 iterations by default.

What is the best way to handle both per-commit and branch-level code reviews?

Handling per-commit and branch-level code reviews requires an automated loop that first resolves per-commit findings, then runs a branch-level review to address any remaining branch-wide issues. This ensures comprehensive continuous quality improvement.

Can I run an automated code review fixer without modifying my working directory?

Yes, running an automated code review fixer without modifying your working directory is possible because the agent operates in an isolated git worktree. This isolated environment keeps your primary working tree completely clean during the iterative fixing process.

What do I need to set up before automating iterative code review fixes?

To automate iterative code review fixes you need a clean working tree within a git repository and the roborev tool installed. These prerequisites ensure the automated agent can safely apply commits and manage the feature branch without conflicts.

Does automated code review fixing stop if issues are not resolved quickly?

Automated code review fixing manages resolution limits by running up to 10 iterations by default to resolve per-commit findings before stopping. You can also adjust this limit using options like --max-iterations and --since to control the scope.

Why does my automated code review loop require a clean working tree?

An automated code review loop requires a clean working tree to ensure the isolated agent can safely commit fixes to the feature branch without overwriting uncommitted local changes. This prevents data loss during the iterative commit process.