code_review

Run Claude Code ultrareview on pull requests or local paths and output structured JSON findings.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/heg-wtf/abyss --skill code-review-heg-wtf
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: code_review
Source: https://github.com/heg-wtf/abyss/tree/main/src/abyss/builtin_skills/code_review
Command: npx skills add https://github.com/heg-wtf/abyss --skill code-review-heg-wtf

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Code reviewers spend time manually examining diffs and summarizing issues; this skill automates that process by running Claude Code ultrareview on a pull request or local target and delivering a structured findings summary.

Core Features & Use Cases

  • Automated code review on PR URLs, PR numbers, or local paths using Claude Code ultrareview.
  • Generates a structured JSON payload with findings and suggested changes for faster decision-making.
  • Use case: a developer wants a quick, reliable review of changes before merging to reduce defects.

Quick Start

Ask Claude Code to ultrareview the given PR URL, PR number, or local path to generate a summarized findings report.

Frequently Asked Questions about code_review

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

FAQPage Schema
What format does the code review output use for downstream automation?▼

The code review output uses a structured JSON payload containing findings and suggested changes with file paths and rationales, designed for direct consumption by downstream automation tools.

How do I automate code review for a GitHub pull request?▼

Automated code review for a GitHub pull request is performed by running Claude Code ultrareview on the PR URL or PR number to generate a structured JSON findings report with file paths and rationales.

Can I review local code changes instead of a pull request?▼

Yes, local code review is supported by passing a local file path to Claude Code ultrareview, which analyzes the diff and outputs a structured JSON payload containing findings and suggested changes.

Does the automated code review work with GitLab and Bitbucket PRs?▼

Yes, automated code review works with GitLab and Bitbucket PRs. The tool applies ultrareview to PR URLs from these platforms and produces a concise findings summary in a structured JSON format.

What is needed to run AI-assisted code review on my repository?▼

AI-assisted code review requires the Claude CLI installed in your environment. It runs the ultrareview command with a default 30-minute timeout to analyze diffs and output a structured JSON payload.

What format does the code review output use for downstream automation?▼

The code review output uses a structured JSON payload containing findings and suggested changes with file paths and rationales, designed for direct consumption by downstream automation tools.