Unity-MCP

#!/usr/bin/env python3 """ MCP Tool JSON Validator using OpenAI API Validates MCP tool JSON by actually injecting it into OpenAI API. Tests if the tool definition is accepted by OpenAI's function calling system. This is a real "battle test" - if OpenAI API accepts it, it's valid! """ import json import sys import os from pathlib import Path # Try to load .env file from script directory try: from dotenv import load_dotenv # Get the directory where this script is located script_dir = Path(__file__).parent env_path = script_dir / '.env' load_dotenv(dotenv_path=env_path) except ImportError: # python-dotenv not installed, will use system environment variables only pass try: from openai import OpenAI except ImportError: print("Error: openai library not found.") print("Install it with: pip install openai") sys.exit(1) # Try to import colorama for Windows color support try: from colorama import init, Fore, Style init(autoreset=True) HAS_COLORAMA = True except ImportError: HAS_COLORAMA = False # ANSI color codes class Colors: if HAS_COLORAMA: GREEN = Fore.GREEN RED = Fore.RED ORANGE = Fore.YELLOW BLUE = Fore.CYAN RESET = Style.RESET_ALL BOLD = Style.BRIGHT else: GREEN = '\033[92m' RED = '\033[91m' ORANGE = '\033[93m' BLUE = '\033[96m' RESET = '\033[0m' BOLD = '\033[1m' def get_openai_client(): """Get OpenAI client with API key from environment or .env file.""" api_key = os.getenv('OPENAI_API_KEY') if not api_key: script_dir = Path(__file__).parent print(f"{Colors.RED}Error: OPENAI_API_KEY not found.{Colors.RESET}") print(f"\nOption 1: Create a .env file in the script directory:") print(f" Location: {script_dir}/.env") print(f" Content: OPENAI_API_KEY=sk-your-api-key-here") print(f"\nOption 2: Set environment variable:") print(f" Linux/Mac: export OPENAI_API_KEY='sk-your-api-key-here'") print(f" Windows: set OPENAI_API_KEY=sk-your-api-key-here") sys.exit(1) return OpenAI(api_key=api_key) def convert_mcp_to_openai_tool(mcp_tool): """ Convert MCP tool definition to OpenAI function calling format. Args: mcp_tool: MCP tool definition from the JSON Returns: OpenAI tool definition """ return { "type": "function", "function": { "name": mcp_tool.get("name", "unknown"), "description": mcp_tool.get("description", ""), "parameters": mcp_tool.get("inputSchema", {}) } } def validate_with_openai(json_content, filename): """ Validate MCP tool JSON by actually injecting it into OpenAI API. Args: json_content: The parsed JSON content filename: Name of the file being validated Returns: Validation result dictionary """ client = get_openai_client() validation_results = { "isValid": True, "errors": [], "warnings": [], "info": [], "summary": "" } try: # Extract tools from MCP JSON tools_list = [] # Check if it's wrapped in a result object (like the example) if "result" in json_content and "tools" in json_content["result"]: mcp_tools = json_content["result"]["tools"] elif "tools" in json_content: mcp_tools = json_content["tools"] elif isinstance(json_content, list): mcp_tools = json_content else: # Assume single tool mcp_tools = [json_content] print(f"{Colors.BLUE}🔍 Injecting {len(mcp_tools)} tool(s) into OpenAI API...{Colors.RESET}") # Convert each MCP tool to OpenAI format for i, mcp_tool in enumerate(mcp_tools): try: openai_tool = convert_mcp_to_openai_tool(mcp_tool) tools_list.append(openai_tool) print(f"{Colors.BLUE} ├─ Tool {i+1}: {openai_tool['function']['name']}{Colors.RESET}") except Exception as e: validation_results["isValid"] = False validation_results["errors"].append({ "severity": "error", "location": f"tools[{i}]", "message": f"Failed to convert MCP tool to OpenAI format: {str(e)}", "suggestion": "Check that the tool has required fields: name, description, inputSchema" }) if not tools_list: validation_results["isValid"] = False validation_results["errors"].append({ "severity": "error", "location": "root", "message": "No valid tools found in the JSON", "suggestion": "Ensure the JSON contains a 'tools' array with tool definitions" }) return validation_results # Try to make a test API call with the tools to validate them print(f"{Colors.BLUE} └─ Testing with OpenAI API...{Colors.RESET}") response = client.chat.completions.create( model="gpt-4o-mini", # Use cheaper model for validation messages=[ { "role": "user", "content": "Hello" } ], tools=tools_list, tool_choice="none" # Don't actually call the tools ) # If we got here, the tools are valid! validation_results["isValid"] = True validation_results["summary"] = f"Successfully validated {len(tools_list)} tool(s) with OpenAI API" validation_results["info"].append({ "severity": "info", "message": f"All {len(tools_list)} tool definition(s) accepted by OpenAI API" }) # Check for missing descriptions or other best practices for i, mcp_tool in enumerate(mcp_tools): tool_name = mcp_tool.get("name", f"tool_{i}") if not mcp_tool.get("description"): validation_results["warnings"].append({ "severity": "warning", "location": f"tools[{i}].description", "message": f"Tool '{tool_name}' is missing a description", "suggestion": "Add a description to help the AI understand when to use this tool" }) return validation_results except Exception as e: # Parse OpenAI API error for specific validation issues error_message = str(e) validation_results["isValid"] = False # Try to extract specific error details if "Invalid schema" in error_message or "invalid" in error_message.lower(): validation_results["errors"].append({ "severity": "error", "location": "inputSchema", "message": f"OpenAI API rejected the schema: {error_message}", "suggestion": "Check that inputSchema follows JSON Schema Draft 7 specification" }) else: validation_results["errors"].append({ "severity": "error", "location": "API call", "message": f"OpenAI API error: {error_message}", "suggestion": "Review the full error message and check your tool definition" }) validation_results["summary"] = "Validation failed - OpenAI API rejected the tool definition" return validation_results def print_validation_results(result, filename): """Print validation results in a colorized, readable format.""" if not result: return False is_valid = result.get('isValid', False) errors = result.get('errors', []) warnings = result.get('warnings', []) info = result.get('info', []) summary = result.get('summary', '') # Print header print("\n" + "="*70) if is_valid: print(f"{Colors.GREEN}{Colors.BOLD}✓ VALIDATION PASSED{Colors.RESET}") else: print(f"{Colors.RED}{Colors.BOLD}✗ VALIDATION FAILED{Colors.RESET}") print("="*70) # Print summary if summary: print(f"\n{Colors.BOLD}Summary:{Colors.RESET}") print(f" {summary}") # Print errors if errors: print(f"\n{Colors.RED}{Colors.BOLD}ERRORS ({len(errors)}):{Colors.RESET}") for i, error in enumerate(errors, 1): print(f"\n {Colors.RED}[{i}] {error.get('message', 'Unknown error')}{Colors.RESET}") if 'location' in error: print(f" {Colors.BOLD}Location:{Colors.RESET} {error['location']}") if 'suggestion' in error: print(f" {Colors.BOLD}Fix:{Colors.RESET} {error['suggestion']}") # Print warnings if warnings: print(f"\n{Colors.ORANGE}{Colors.BOLD}WARNINGS ({len(warnings)}):{Colors.RESET}") for i, warning in enumerate(warnings, 1): print(f"\n {Colors.ORANGE}[{i}] {warning.get('message', 'Unknown warning')}{Colors.RESET}") if 'location' in warning: print(f" {Colors.BOLD}Location:{Colors.RESET} {warning['location']}") if 'suggestion' in warning: print(f" {Colors.BOLD}Recommendation:{Colors.RESET} {warning['suggestion']}") # Print info if info: print(f"\n{Colors.BLUE}{Colors.BOLD}INFORMATION ({len(info)}):{Colors.RESET}") for i, item in enumerate(info, 1): print(f" {Colors.BLUE}• {item.get('message', '')}{Colors.RESET}") # Print footer print("\n" + "="*70) print(f"File: {filename}") print(f"Status: {Colors.GREEN}VALID{Colors.RESET}" if is_valid else f"Status: {Colors.RED}INVALID{Colors.RESET}") if errors or warnings: print(f"Issues: {Colors.RED}{len(errors)} error(s){Colors.RESET}, {Colors.ORANGE}{len(warnings)} warning(s){Colors.RESET}") print("="*70 + "\n") return is_valid def validate_mcp_tool_file(filename): """ Validate an MCP tool JSON file. Args: filename: Path to the MCP tool JSON file Returns: True if valid, False otherwise """ file_path = Path(filename) # Check if file exists if not file_path.exists(): print(f"{Colors.RED}Error: File '{filename}' not found.{Colors.RESET}") return False # Read and parse the JSON file try: with open(file_path, 'r', encoding='utf-8') as f: json_content = json.load(f) except json.JSONDecodeError as e: print(f"{Colors.RED}Error: Invalid JSON format in '{filename}'{Colors.RESET}") print(f" Line {e.lineno}, Column {e.colno}: {e.msg}") return False except Exception as e: print(f"{Colors.RED}Error reading file: {e}{Colors.RESET}") return False print(f"{Colors.BLUE}Validating MCP Tool JSON: {filename}{Colors.RESET}") # Validate using OpenAI result = validate_with_openai(json_content, filename) if result: return print_validation_results(result, filename) else: return False def main(): """Main function to handle command line arguments and user input.""" filename = None # Check if filename was provided as command line argument if len(sys.argv) > 1: filename = sys.argv[1] else: # Prompt user for filename print(f"{Colors.BOLD}MCP Tool JSON Validator (OpenAI-powered){Colors.RESET}") print("-" * 50) filename = input("Enter the MCP tool JSON filename: ").strip() if not filename: print(f"{Colors.RED}Error: No filename provided.{Colors.RESET}") sys.exit(1) # Validate the MCP tool JSON success = validate_mcp_tool_file(filename) # Exit with appropriate status code sys.exit(0 if success else 1) if __name__ == "__main__": main()