MCP Simple OpenAI Assistant

"""Tool definitions for the OpenAI Assistant MCP server.""" from mcp.types import Tool def get_tool_definitions() -> list[Tool]: """Get all tool definitions for the server.""" return [ Tool( name="create_assistant", description="""Create a new OpenAI assistant to help you with your tasks, you can provide instructions that this assistant will follow when working with your prompts and you can provide a model to use for the assistant. NOTE: It is recomended to check existing assistants with list_assistants before creating a new one. This will help you avoid creating assistants with the same name or instructions. """, inputSchema={ "type": "object", "properties": { "name": { "type": "string", "description": """Name for the assistant, use a descriptive name to be able to re-use it in the future. Remember, that your other instance will not have the context you have, make sure the assistant name is descriptive enough. """ }, "instructions": { "type": "string", "description": """Instructions for the assistant that will shape its behavior and responses. These can be quite long if you feel this will help the assistant to perform better. """ }, "model": { "type": "string", "description": "Model to use (default: gpt-4o)", "default": "gpt-4o" } }, "required": ["name", "instructions"] } ), Tool( name="new_thread", description="Creates a new conversation thread. Threads have large capacity and the context window is moving so that it always covers a certain number of tokens (depending on the model).", inputSchema={ "type": "object", "properties": {}, "additionalProperties": False } ), Tool( name="send_message", description="""Send a message to assistant and start processing. The response will not be immediately available - use check_response with the same thread_id to get it when ready.""", inputSchema={ "type": "object", "properties": { "thread_id": { "type": "string", "description": "Thread ID to use" }, "assistant_id": { "type": "string", "description": "Assistant ID to use" }, "message": { "type": "string", "description": "Message to send" } }, "required": ["thread_id", "assistant_id", "message"] } ), Tool( name="check_response", description="""Check if assistant's response is ready in the thread. Returns either 'in_progress' status or the actual response if ready.""", inputSchema={ "type": "object", "properties": { "thread_id": { "type": "string", "description": "Thread ID to check" } }, "required": ["thread_id"] } ), Tool( name="list_assistants", description="""List all available OpenAI assistants. Returns a list of assistants with their IDs, names, and configurations. Use this to find existing assistants you can work with. The results can be used with other tools like send_message or update_assistant.""", inputSchema={ "type": "object", "properties": { "limit": { "type": "number", "description": "Optional: Maximum number of assistants to return (default: 20)", "default": 20 } }, "additionalProperties": False } ), Tool( name="retrieve_assistant", description="Get details of a specific assistant", inputSchema={ "type": "object", "properties": { "assistant_id": { "type": "string", "description": "ID of the assistant to retrieve" } }, "required": ["assistant_id"] } ), Tool( name="update_assistant", description="Modify an existing assistant", inputSchema={ "type": "object", "properties": { "assistant_id": { "type": "string", "description": "ID of the assistant to modify" }, "name": { "type": "string", "description": "Optional: New name for the assistant" }, "instructions": { "type": "string", "description": "Optional: New instructions for the assistant" }, "model": { "type": "string", "description": "Optional: New model to use (e.g. gpt-4o)" } }, "required": ["assistant_id"] } ) ]