Skip to content

AGENTS.md

Latest commit

 

History

History
73 lines (56 loc) · 2.23 KB

AGENTS.md

File metadata and controls

73 lines (56 loc) · 2.23 KB

Claude Code Review - Agent Guidelines

Project Overview

AI-powered code review tool using Claude to analyze PRs for code quality and security issues. Uses a unified multi-agent approach for comprehensive analysis in a single pass.

Architecture

claudecode/
├── github_action_audit.py   # Main orchestrator - entry point
├── prompts.py               # Review prompt templates
├── findings_filter.py       # False positive filtering
├── claude_api_client.py     # Claude API client
├── json_parser.py           # JSON extraction utilities
├── constants.py             # Configuration constants
└── evals/                   # Evaluation framework

Key Patterns

Prompt Structure

  • get_unified_review_prompt() - Combined code quality + security review
  • Prompts require JSON-only output with specific schema
  • Always include confidence scores (0.7-1.0 threshold)

Finding Categories

  • Code Quality: correctness, reliability, performance, maintainability, testing
  • Security: security (injection, auth, crypto, data exposure)

Severity Levels

  • HIGH: Production bugs, data loss, exploitable vulnerabilities
  • MEDIUM: Limited scope, specific conditions required
  • LOW: Minor issues, use sparingly

Filtering Pipeline

  1. Hard exclusion rules (regex patterns in HardExclusionRules)
  2. Claude API validation (optional, uses claude_api_client.py)
  3. Directory exclusion filtering

Testing

# Python tests
pytest claudecode -v  # Run all tests (177 passing)
# JavaScript tests
~/.bun/bin/bun test scripts/comment-pr-findings.bun.test.js

Code Style

  • Python 3.9+
  • Type hints encouraged
  • Comprehensive docstrings
  • Tests required for new functionality

Key Files

File Purpose
action.yml GitHub Action definition
.claude/commands/review.md Slash command for Claude Code
docs/ Customization documentation

Common Tasks

Adding a review category

  1. Update category lists in prompts.py
  2. Add exclusion patterns in findings_filter.py if needed
  3. Add tests in test_prompts.py

Modifying filtering rules

  1. Edit HardExclusionRules in findings_filter.py
  2. Add tests in test_hard_exclusion_rules.py