Code Quality Checker for Claude Code

code-quality-checker.py

#!/usr/bin/env python3
import json
import os
import re
import sys


def check_openai_models(content: str) -> list[str]:
    """Check for outdated OpenAI model usage."""
    issues = []

    # Pattern for OpenAI model references
    openai_patterns = [
        (r'gpt-4(?:-turbo)?(?:-preview)?(?!\.1|5)', 'Use latest OpenAI models: gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, or reasoning models gpt-5, gpt-5-mini, gpt-5-nano instead of older GPT-4 variants'),
        (r'gpt-3\.5-turbo', 'Use latest OpenAI models: gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, or reasoning models gpt-5, gpt-5-mini, gpt-5-nano instead of GPT-3.5'),
        (r'text-davinci-003', 'Use latest OpenAI models: gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, or reasoning models gpt-5, gpt-5-mini, gpt-5-nano instead of legacy models'),
    ]

    for pattern, message in openai_patterns:
        if re.search(pattern, content, re.IGNORECASE):
            issues.append(message)

    # Check for temperature and max_tokens usage with OpenAI
    if re.search(r'temperature\s*[:=]|max_tokens\s*[:=]', content) and re.search(r'gpt-|openai', content, re.IGNORECASE):
        issues.append('Do not pass temperature or max_tokens parameters with OpenAI models - use default settings')

    return issues

def check_anthropic_models(content: str) -> list[str]:
    """Check for outdated Anthropic model usage."""
    issues = []

    # Pattern for outdated Claude models
    anthropic_patterns = [
        (r'claude-3-5-sonnet|claude-3\.5-sonnet', 'Use claude-4-sonnet (Sonnet 4.0) instead of Claude 3.5 Sonnet'),
        (r'claude-3-7-sonnet|claude-3\.7-sonnet', 'Use claude-4-sonnet (Sonnet 4.0) instead of Claude 3.7 Sonnet'),
        (r'claude-3-opus', 'Use claude-4-sonnet (Sonnet 4.0) for best performance'),
        (r'claude-3-haiku', 'Use claude-4-sonnet (Sonnet 4.0) for better capabilities'),
    ]

    for pattern, message in anthropic_patterns:
        if re.search(pattern, content, re.IGNORECASE):
            issues.append(message)

    return issues

def check_fallback_patterns(content: str) -> list[str]:
    """Check for fallback and legacy patterns."""
    issues = []

    fallback_patterns = [
        (r'\bfallback\b', 'WARNING: Fallbacks detected. Code should fail fast and throw errors early. Consider rewriting without fallback logic.'),
        (r'\blegacy\b', 'WARNING: Legacy code detected. Consider removing legacy code and using modern implementations.'),
        (r'backward[s]?\s+compatib', 'WARNING: Backwards compatibility detected. Consider breaking existing code if needed for better implementation.'),
        (r'\|\|\s*[\'"][^\'"]*[\'"]', 'WARNING: Default value fallbacks (|| "default") detected. Consider using explicit error handling instead.'),
        (r'\?\?\s*[\'"][^\'"]*[\'"]', 'WARNING: Nullish coalescing with defaults (?? "default") detected. Consider replacing with explicit validation.'),
        (r'try\s*\{[^}]*\}\s*catch\s*\([^)]*\)\s*\{[^}]*\}', 'WARNING: Empty or generic catch blocks may hide errors. Ensure proper error handling and re-throwing.'),
    ]

    for pattern, message in fallback_patterns:
        if re.search(pattern, content, re.IGNORECASE):
            issues.append(message)

    return issues

def check_any_type_usage(content: str) -> list[str]:
    """Check for 'any' type usage in TypeScript/JavaScript."""
    issues = []

    # More sophisticated pattern to catch 'any' as a type
    any_type_patterns = [
        (r':\s*any\b', "WARNING: 'any' type detected. Consider defining proper types instead. Look up existing type definitions or create new ones in appropriate locations."),
        (r'<any>', "WARNING: 'any' generic type detected. Consider using specific types instead."),
        (r'Array<any>', "WARNING: 'Array<any>' detected. Consider using 'Array<SpecificType>' instead."),
        (r'Promise<any>', "WARNING: 'Promise<any>' detected. Consider using 'Promise<SpecificType>' instead."),
        (r'\bas\s+any\b', "WARNING: Type assertion 'as any' detected. Consider using proper type assertions or fix the underlying type issue."),
        (r'Record<[^,>]+,\s*any>', "WARNING: 'Record<string, any>' pattern detected. Consider defining proper value types if possible."),
    ]

    # Don't match 'any' in strings, comments, or variable names
    # Split content into lines and check each line
    lines = content.split('\n')
    for line_num, line in enumerate(lines, 1):
        # Skip comments
        if re.match(r'^\s*(/\*|//|\*|#)', line.strip()):
            continue

        # Skip string literals (simplified - may not catch all cases)
        line_without_strings = re.sub(r'["\'][^"\']*["\']', '', line)
        line_without_strings = re.sub(r'`[^`]*`', '', line_without_strings)

        for pattern, message in any_type_patterns:
            if re.search(pattern, line_without_strings):
                issues.append(f"{message} (Line {line_num})")
                break  # Only report one issue per line

    return issues

def main():
    try:
        input_data = json.load(sys.stdin)
    except json.JSONDecodeError as e:
        print(f"Error: Invalid JSON input: {e}", file=sys.stderr)
        sys.exit(1)

    tool_name = input_data.get("tool_name", "")
    tool_input = input_data.get("tool_input", {})
    hook_event = input_data.get("hook_event_name", "PreToolUse")

    # Only check Write and Edit operations
    if tool_name not in ["Write", "Edit", "MultiEdit"]:
        sys.exit(0)

    # Get file content
    content = ""
    if tool_name == "Write":
        content = tool_input.get("content", "")
    elif tool_name == "Edit":
        content = tool_input.get("new_string", "")
    elif tool_name == "MultiEdit":
        # Check all edits
        edits = tool_input.get("edits", [])
        content = "\n".join([edit.get("new_string", "") for edit in edits])

    if not content:
        sys.exit(0)

    # Get file path for context
    file_path = tool_input.get("file_path", "")

    # Collect blocking issues (model-related)
    blocking_issues = []
    blocking_issues.extend(check_openai_models(content))
    blocking_issues.extend(check_anthropic_models(content))

    # Collect warning issues
    warning_issues = []
    warning_issues.extend(check_fallback_patterns(content))
    warning_issues.extend(check_any_type_usage(content))

    # Handle blocking issues
    if blocking_issues:
        error_message = f"Code quality issues detected in {file_path}:\n\n"
        for i, issue in enumerate(blocking_issues, 1):
            error_message += f"{i}. {issue}\n"

        error_message += "\nPlease address these issues and try again."

        # Use JSON output to provide structured feedback
        output = {
            "decision": "block",
            "reason": error_message
        }
        print(json.dumps(output))
        sys.exit(0)

    # Handle warnings (show but don't block)
    if warning_issues:
        warning_message = f"Code quality issues detected in {file_path}:\n\n"
        for i, issue in enumerate(warning_issues, 1):
            warning_message += f"{i}. {issue}\n"

        warning_message = f"""<system-reminder>

{warning_message}

It is strongly advised to address these issues and try again.
</system-reminder>"""

        # Use JSON output to show warnings without blocking
        output = {
            "hookSpecificOutput": {
                "hookEventName": hook_event,
                "additionalContext": warning_message
            }
        }
        print(json.dumps(output))
        sys.exit(0)

    # If no issues, allow the operation to proceed
    sys.exit(0)

if __name__ == "__main__":
    main()

I did not create this and don't remember where I ripped it from. But it works!

The hook:

    "PostToolUse": [
      {
        "matcher": "Edit|MultiEdit|Write",
        "hooks": [
          {
            "type": "command",
            "command": "uv run .claude/hooks/code-quality-checker.py"
          }
        ]
      }
    ]