GitHub PR Reviewer
Overview
Automatically analyze pull requests and post actionable inline code review comments covering security, performance, style, and correctness using AI-assisted static analysis.
When to Use
- •Auto-reviewing PRs as part of a CI/CD pipeline
- •Enforcing code standards when human reviewers are unavailable
- •Catching security vulnerabilities before they reach main
- •Ensuring test coverage meets minimum thresholds
- •Generating first-pass review comments to speed up human review
Instructions
- •Accept inputs: repository name, PR number, GitHub token, review strictness level (light/standard/strict).
- •Fetch the PR diff and file list via GitHub REST API.
- •Analyze each changed file for: syntax errors, unused imports, security vulnerabilities (SQL injection, hardcoded secrets, XSS), missing error handling, test coverage gaps.
- •Check compliance with project's existing coding patterns from the codebase.
- •Generate inline review comments with line numbers, issue descriptions, and suggested fixes.
- •Summarize findings: total issues, breakdown by severity, overall assessment (Approve/Request Changes/Comment).
- •Post the review via GitHub API (pulls.createReview) with inline comments.
- •If no critical issues: approve with notes. If critical issues found: request changes.
Environment
code
GITHUB_TOKEN=ghp_your_personal_access_token GITHUB_REPO=owner/repository REVIEW_STRICTNESS=standard POST_REVIEW=true SECURITY_SCAN=true
Examples
Input:
code
repo: Greenmamba29/skillsdotmd_web pr_number: 42 strictness: strict auto_approve_if_clean: true
Output:
code
PR #42 Review Complete Verdict: Request Changes Critical: 1 (hardcoded API key on line 23) High: 2 (missing error handling in auth.js) Medium: 4 (unused imports, style violations) Comments posted: 7 inline Summary review posted to GitHub