Code Review Automation.prompt

# Automated Code Review with LLMs

## Overview

Automated code review uses LLMs to analyze code changes, identify bugs, security vulnerabilities, performance issues, and style violations before human review. Unlike static analysis tools that follow rigid rules, LLM-based review understands code semantics, can reason about edge cases, and provides natural language explanations. This augments human reviewers by catching issues that rule-based tools miss.

Production code review automation typically runs as a CI/CD check on pull requests, generating review comments posted directly on the PR.

## Key Concepts

- **Diff-aware review**: Focus the LLM on changed lines and their surrounding context
- **Multi-pass review**: Run multiple specialized prompts (security, performance, style) in parallel
- **Severity classification**: Categorize findings as critical, warning, or suggestion
- **False positive filtering**: Use a second LLM pass to filter out likely false positives
- **Context enrichment**: Include file history, related PRs, and project conventions
- **Review templates**: Standardized prompts for different review aspects

## Implementation Patterns