Atlas / MCP servers / xncbf / ChatGPT

ChatGPTSAFE

mcp/xncbf/chatgpt-2

MCP server enabling interaction between macOS ChatGPT desktop app and MCP-compatible AI assistants.

Verdict
SAFE
Grade
B
Trust score
89 /100
Exposed tools
3 1r · 2w · 0d
Transport
stdio
License
MIT
Stars
80
01

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

A Model Context Protocol (MCP) server that enables AI assistants to interact with the ChatGPT desktop app on macOS.

https://github.com/user-attachments/assets/a30c9b34-cdbe-4c0e-a0b0-33eb5054db5c

Language Support

Supported system languages for response detection:

  • Korean
  • English

If your macOS system language is not listed above, please follow these instructions:

  1. Make sure ChatGPT desktop app is running
  2. Run show_all_button_names.applescript and copy the output to create an issue for language support.

Features

  • Send prompts to ChatGPT from any MCP-compatible AI assistant
  • Built with Python and FastMCP

Note: This server only supports English text input. Non-English characters may not work properly.

Installation

Prerequisites

  • macOS
  • ChatGPT desktop app installed and running
  • Python 3.10+
  • uv package manager

For Claude Code Users

Simply run:

claude mcp add chatgpt-mcp uvx chatgpt-mcp

That's it! You can start using ChatGPT commands in Claude Code.

For Other MCP Clients

Step 1: Install the MCP Server

Option A: Install from PyPI (Recommended)

# Install with uv
uv add chatgpt-mcp

Option B: Manual Installation

# Clone the repository
git clone https://github.com/xncbf/chatgpt-mcp
cd chatgpt-mcp

# Install dependencies with uv
uv sync

Step 2: Configure Your MCP Client

If installed from PyPI, add to your MCP client configuration:

{
"mcpServers": {
"chatgpt": {
"command": "uvx",
"args": ["chatgpt-mcp"]
}
}
}

If manually installed, add to your MCP client configuration:

{
"mcpServers": {
"chatgpt": {
"command": "uv",
"args": ["run", "chatgpt-mcp"],
"cwd": "/path/to/chatgpt-mcp"
}
Read from source at commit e6ddd34e72aaOBSERVED · 2026-10-07
02

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.

claude-code
claude mcp add chatgpt-mcp -- uvx chatgpt-mcp
claude-desktop
{
  "mcpServers": {
    "chatgpt-mcp": {
      "command": "uvx",
      "args": [
        "chatgpt-mcp"
      ]
    }
  }
}
03

Exposed tools (3)

1 read · 2 write · 0 destructive.

ToolRiskDescription
ask_chatgpt_toolwriteSend a prompt to ChatGPT and return the response.
get_chatgpt_response_toolreadGet the latest response from ChatGPT after sending a message.
new_chatgpt_chat_toolwriteStart a new chat conversation in ChatGPT.
04

Trust audit

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

LayerWhat it checksResult
L0Provenance & inventoryPASS
L1Static analysis of the codePASS
L2Instruction surface (what it tells the agent)PASS
L3Class-specific surfacePASS
L4Behavioural (sandbox)SKIPPED

What the source does

Filesystem
none-observed
Network
none-observed
Shell
none-observed
Dependencies
pinned
Secrets in source
none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.

Audited 2026-10-07 · audit v0.4.1 · source sha e6ddd34e72aafull audit observations/trust-audit/mcp-server/xncbf__chatgpt-2.json · Report an issue / request a re-scan
05

Audit history

Every audit this server has had. A grade with a past is a grade somebody is still checking.

DateSourceVerdictGradeScoreChange
2026-10-07e6ddd34e72aaSAFEB89first audit
06

Questions

What is the ChatGPT MCP server?

MCP server enabling interaction between macOS ChatGPT desktop app and MCP-compatible AI assistants.

What tools does ChatGPT expose?

3 in total: 1 read-only, 2 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.

Is ChatGPT safe to connect to an agent?

The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.

What credentials does ChatGPT need?

No credential environment variables were found in its source, so it appears to need none.

How does ChatGPT run?

It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as chatgpt-mcp.

How current is this page?

The grade is for one exact copy of the source (e6ddd34e72aa), read on 2026-10-07. The repository is watched and re-audited when it changes.

Advertisement