Stats
Actions
Tags
From review-agent
Review a GitHub pull request for code quality issues and refactoring opportunities. Orchestrates code-reviewer and refactoring-advisor agents. Usage: /review-pr <pr-number-or-url> [--no-post] [--review-only] [--refactor-only]
How this skill is triggered — by the user, by Claude, or both
Slash command
/review-agent:review-prThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are orchestrating a comprehensive code review of a pull request. You will coordinate specialized agents to analyze code quality and suggest improvements.
You are orchestrating a comprehensive code review of a pull request. You will coordinate specialized agents to analyze code quality and suggest improvements.
The user invoked this skill with: $ARGUMENTS
Parse the arguments:
42) or full GitHub URL (e.g., https://github.com/owner/repo/pull/42)--no-post: Skip posting the review to GitHub; display the report only--review-only: Only run the code review agent, skip refactoring--refactor-only: Only run the refactoring agent, skip code reviewExtract the PR number from the arguments. If a URL was provided, parse out the owner, repo, and PR number.
If only a number was provided, detect the current repository:
gh repo view --json owner,name -q '.owner.login + "/" + .name'
Then validate the PR exists and is open:
gh pr view <number> --json state,isDraft,title,author,url,number,headRefName,baseRefName
If the PR is closed or merged, inform the user and stop. If the PR is a draft, warn but continue.
Display a summary:
Reviewing PR #<number>: <title>
Author: <author> | Branch: <head> → <base> | State: <state>
Fetch the list of changed files to understand the PR scope:
gh pr view <number> --json files -q '.files[].path'
Report the scope:
Files changed: <count>
<list of filenames>
Launch the specialized agents based on the flags provided.
Default (no flags) or both needed: Launch BOTH agents in parallel using the Agent tool:
code-reviewer agent: "Review PR # in / for bugs, security issues, and style violations. The PR changes these files: . Use the MCP tools mcp__plugin_review-agent_review-agent__get_pr_diff and mcp__plugin_review-agent_review-agent__get_pr_files to fetch the diff. Read the review://standards resource for coding standards."
refactoring-advisor agent: "Analyze PR # in / for refactoring opportunities. The PR changes these files: . Use the MCP tools mcp__plugin_review-agent_review-agent__get_pr_diff and mcp__plugin_review-agent_review-agent__get_pr_files to fetch the diff. Focus on structural improvements that preserve behavior."
With --review-only: Launch only the code-reviewer agent.
With --refactor-only: Launch only the refactoring-advisor agent.
If an agent fails (e.g., MCP tool error, rate limit, network issue): include an error notice in that section of the report ("Code review failed: {error message}") and continue with the other agent's results. Do not abort the entire review because one agent failed.
Once all launched agents complete (or fail), compile their findings into a unified report:
# PR Review Report: #<number> — <title>
## Overview
- **Author**: <author>
- **Branch**: <head> → <base>
- **Files Changed**: <count>
- **Review Date**: <current date>
---
## Code Review Findings
<Insert code-reviewer agent results here>
---
## Refactoring Suggestions
<Insert refactoring-advisor agent results here>
---
## Final Recommendation
<Based on the findings:>
<- If critical issues found: **REQUEST_CHANGES** with explanation>
<- If only suggestions: **COMMENT** with summary>
<- If code is clean: **APPROVE** with praise>
Use the mcp__plugin_review-agent_review-agent__post_review_comment MCP tool to post the review:
body to the full aggregated review reportevent to:
REQUEST_CHANGES if critical issues were foundCOMMENT if only non-critical findingsAPPROVE if no issues foundcomments array for each finding that has a specific file:line referenceIf the --no-post flag was provided, skip this step and only display the report.
npx claudepluginhub siarhei-prystauka/review-agent --plugin review-agentProvides UI/UX resources: 50+ styles, color palettes, font pairings, guidelines, charts for web/mobile across React, Next.js, Vue, Svelte, Tailwind, React Native, Flutter. Aids planning, building, reviewing interfaces.
Searches MemPalace before answering questions about past work, people, projects, or prior decisions. Returns verbatim stored content instead of guessing from model memory.