Model Context Protocol moderate risk

Simple OpenAI Assistant

A simple MCP server that enables interaction with OpenAI assistants, providing tools to manage assistants and threads, and stream responses.

Connections & Capabilities

Connects To

GitHubSQLite

Capabilities

readwriteexecadmin

Quickstart

Install

npx -y @smithery/cli install mcp-simple-openai-assistant --client claude

Config

{
  "mcpServers": {
    "openai-assistant": {
      "command": "python",
      "args": ["-m", "mcp_simple_openai_assistant"],
      "env": {
        "OPENAI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Exposed MCP Tools (8)

moderate

create_assistant

Creates a new OpenAI assistant with specified name, instructions, and model.

Allows creation of new assistants, potentially consuming resources.

safe

list_assistants

Lists all available OpenAI assistants associated with the API key.

Read-only operation, no side effects.

safe

retrieve_assistant

Retrieves detailed information about a specific OpenAI assistant.

Read-only operation, no side effects.

high

update_assistant

Modifies an existing OpenAI assistant's name, instructions, or model.

Allows modification of assistant configurations, potentially impacting behavior.

moderate

create_new_assistant_thread

Creates a new conversation thread with a user-defined name and description.

Creates new threads, potentially consuming resources.

safe

list_threads

Lists all locally managed conversation threads from the database.

Read-only operation, no side effects.

critical

delete_thread

Deletes a conversation thread from both OpenAI's servers and the local database.

Deletes threads, resulting in permanent data loss.

moderate

ask_assistant_in_thread

Sends a message to an assistant within a thread and streams the response back in real-time.

Sends messages and receives responses, potentially triggering unintended actions.

Safety Assessment

This server provides useful tools for managing OpenAI assistants, but its safety relies heavily on the security of the OpenAI API key. It is safe to use within a controlled environment where the API key is protected. Risks increase if the API key is exposed or if the server is used with untrusted inputs.

Requires OpenAI API key, limiting access to authorized users.
Provides tools for managing assistants and threads, not direct code execution.
Local persistence of threads improves management without exposing sensitive data externally.
Streaming responses enhance user experience and reduce timeout risks.

Relies on the security of the OpenAI API key, which if compromised, grants full access.
Allows creation, updating, and deletion of assistants and threads.
No explicit input validation or sanitization mentioned, potentially leading to prompt injection.
Manages threads locally, which could be a vulnerability if the database is compromised.
Lacks fine-grained access controls; access is all-or-nothing based on the API key.