pr-review-loop

Iteratively review GitHub PRs until zero unresolved threads and green checks.

Updated May 11, 2026
One-click install
npx skills add https://github.com/jason660519/Project-Manager --skill pr-review-loop-jason660519
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pr-review-loop
Source: https://github.com/jason660519/Project-Manager/tree/main/.agents/skills/pr-review-loop
Command: npx skills add https://github.com/jason660519/Project-Manager --skill pr-review-loop-jason660519

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Iteratively drive an open GitHub PR for Project Manager to a clean review — loop of wait for bot review on HEAD → fix actionable findings → verify baseline → push → reply and resolve threads → re-trigger review, until zero unresolved review threads and green checks, or a max-iteration cap. The PM-specific replacement for greptileai/skills/greploop (this repo has no Greptile; the reviewers are chatgpt-codex-connector, copilot-pull-request-reviewer, GitGuardian, and the verify CI). Use when the user says loop the PR / keep fixing review comments until clean / drive the bots to zero / converge the PR / greploop. Builds on the check-pr skill (which does one pass); this one loops. Does NOT merge — that's ship.

Core Features & Use Cases

  • Loops through PR review steps: wait for bot review on HEAD, fix findings, verify baseline, push, reply and resolve threads, and re-trigger reviews, until all checks pass or cap reached.
  • Keeps PR workflow deterministic by tracking iteration counts and finding fingerprints to avoid churn.
  • Explicitly handles outdated threads by ensuring convergence targets include only current checks and updated HEAD state, without merging.

Quick Start

Run this skill against an open PR to start looping the review process until convergence.

Frequently Asked Questions about pr-review-loop

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

FAQPage Schema
How do I automate iterative GitHub PR reviews until all checks pass?

Automated iterative PR review looping drives an open GitHub pull request to zero unresolved threads and green checks by repeatedly waiting for bot reviews, fixing findings, pushing commits, and resolving threads. It applies deterministic looping with an iteration cap.

Why does my pull request review loop keep churning through the same bot comments?

PR review loop churn is prevented by tracking iteration counts and finding fingerprints. This ensures the convergence targets only current checks and the updated HEAD state, avoiding repetitive fixes for outdated threads during the automated review process.

Can I use automated PR looping to converge CodeX, Copilot, and GitGuardian bot reviews?

Automated PR looping supports converging reviews from chatgpt-codex-connector, copilot-pull-request-reviewer, GitGuardian, and verify CI checks. It iteratively fixes actionable findings from these bot reviewers until all threads are resolved and checks are green.

What is the best way to drive an open PR to zero unresolved review threads without merging?

Driving an open PR to zero unresolved threads without merging requires an iterative review loop that fixes findings, verifies the baseline, pushes, and resolves threads. It explicitly handles outdated threads and stops at convergence without performing the merge action.

Does automated PR looping handle CI check failures and outdated review threads?

Automated PR looping explicitly handles CI check failures and outdated review threads by verifying the baseline, ensuring convergence targets include only current checks, and updating the HEAD state. It stops when all checks are green or a max-iteration cap is reached.

How do I start looping the PR review process for an open pull request?

To start looping the PR review process, run the skill against an open PR. It will iteratively wait for bot reviews on HEAD, fix actionable findings, verify the baseline, push commits, reply to threads, and re-trigger reviews until convergence or the iteration cap.