ralpr

Convert GitHub issues into pull requests via three-phase iterative workflows.

Updated Jan 29, 2026
One-click install
npx skills add https://github.com/Archethect/ralpr --skill ralpr
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralpr
Source: https://github.com/Archethect/ralpr/tree/main/skills/ralpr
Command: npx skills add https://github.com/Archethect/ralpr --skill ralpr

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralpr automates the iterative work required to convert GitHub issues into merge-ready pull requests by running a controlled three-phase loop (Implementation → Review → Refactor) and raising cumulative confidence until merge criteria are met.

Core Features & Use Cases

  • Three-phase loop: Run a single, sequential iteration of Implementation, Review, or Refactor to make incremental, reviewable progress on an issue or PR.
  • Parallel AI reviewers: Aggregate feedback from QA, Domain Expert, and Codex reviewers and apply fix-by-default rules until confidence thresholds are reached.
  • Operational safety: Enforces worktree isolation, documentation-first authority rules, structured state stored in PR comments, and label-based tracking to avoid collisions and drift.
  • Use Case: Automatically take a reported bug, implement a test-driven fix, run parallel AI reviewers to identify further fixes, and iterate until the PR meets repository confidence labels for merging.

Quick Start

Run the ralpr skill to process issue 123 through a single implementation iteration by providing the issue number and desired phase to the skill.

Frequently Asked Questions about ralpr

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

FAQPage Schema
How do I automate converting GitHub issues into merge-ready pull requests?

Automating GitHub issue to pull request conversion involves running a three-phase loop: Implementation, Review, and Refactor. This process manages worktrees, applies AI reviewer feedback, and tracks progress via PR comments until merge confidence thresholds are reached.

What is the three-phase iteration loop for AI code review?

The three-phase iteration loop for AI code review consists of sequential Implementation, Review, and Refactor stages. It aggregates feedback from parallel AI reviewers like QA and Domain Experts, applying fixes until the pull request reaches required confidence labels for merging.

Can I run a single implementation or review phase instead of a full loop?

Yes, you can run a single iteration of the Implementation, Review, or Refactor phase independently. This allows you to make incremental, reviewable progress on a specific issue or pull request by passing the issue number and desired phase to the interface.

Does this automated pull request workflow use Git worktrees for isolation?

Yes, the workflow enforces Git worktree isolation to prevent collisions and state drift. This operational safety measure ensures that iterative AI reviews and code changes remain structured and separated throughout the implementation, review, and refactor phases.

How do parallel AI reviewers improve pull request confidence?

Parallel AI reviewers improve pull request confidence by aggregating feedback from QA, Domain Expert, and Codex reviewers. The system applies fix-by-default rules based on this feedback during the Refactor phase until repository confidence labels and merge criteria are satisfied.

What are the limitations of using AI to automate pull request reviews?

A limitation of automating pull request reviews with AI is the strict reliance on documentation-first authority rules and structured state stored in PR comments. The process requires label-based tracking and may not suit workflows unable to enforce worktree isolation or predefined confidence thresholds.