ChatGPT MCP Bridge
Controls an existing Google Chrome session via chrome-devtools-mcp, enabling navigation, tab management, clicking, typing, screenshots, DevTools inspection, and optional coordinate/vision tools.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@ChatGPT MCP Bridgelist files in my Downloads folder"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
ChatGPT MCP Bridge
Use ChatGPT with the files, terminal, and logged-in Chrome session already on your own computer — through OpenAI Secure MCP Tunnel, without exposing a public MCP endpoint.
Website · Latest Release · Glama · Security · Issues

ChatGPT MCP Bridge is a cross-platform desktop/tray app for Linux and Windows. It gives ChatGPT two independent local capabilities:
Connection | What it does |
Local MCP | Browse/search approved folders, read files, optionally create/edit files, and optionally run terminal commands. |
Web MCP | Control your existing Chrome session with |
Your browser session, local files, and shell stay on your machine. ChatGPT reaches the bridge through OpenAI Secure MCP Tunnel.
Important: ChatGPT MCP Bridge does not create OpenAI tunnels or API keys for you. You create them in your own OpenAI Platform account and enter them into the app.
Why this exists
Most MCP servers are hosted somewhere else. That is not useful when the thing you want ChatGPT to work with is already on your own PC — your project folders, terminal, development tools, and authenticated Chrome tabs.
ChatGPT MCP Bridge turns that local environment into controlled MCP capabilities while keeping the runtime local.
Typical uses include:
ask ChatGPT to inspect or update a local project without repeatedly uploading files;
let ChatGPT run approved development commands such as
git status, builds, tests, or ADB commands;work with websites already open and logged in inside your normal Chrome profile;
inspect browser console/network output while debugging;
keep Local MCP and Web MCP independently enabled or disabled;
expose only the folders and capabilities you actually want to use.
Related MCP server: chatgpt-local-coding-agent
Architecture
OpenAI Secure MCP Tunnel
|
+--------------------+--------------------+
| |
Local Files app Chrome app
| |
v v
ChatGPT MCP Bridge ---------------------- ChatGPT MCP Bridge
| |
Local MCP Chrome DevTools MCP
/ | \ |
files folders optional shell existing ChromeThere is no need to expose a local MCP port directly to the public internet.
Quick start
1. Install the bridge
Linux
Current packages target Debian / Ubuntu / Linux Mint on amd64/x86_64.
curl -fsSL -H "Accept: application/vnd.github.raw+json" "https://api.github.com/repos/haseebgb92/ChatGPT-MCp/contents/install-linux.sh?ref=main" | bashIf you already have Chrome/Chromium and do not want the installer to install Google Chrome:
curl -fsSL -H "Accept: application/vnd.github.raw+json" "https://api.github.com/repos/haseebgb92/ChatGPT-MCp/contents/install-linux.sh?ref=main" | SKIP_CHROME=1 bashOr download the latest .deb manually from GitHub Releases:
sudo apt install ./ChatGPT-MCP-Bridge*.debWindows
Open PowerShell:
Set-ExecutionPolicy -Scope Process Bypass -Force; irm "https://raw.githubusercontent.com/haseebgb92/ChatGPT-MCp/main/install-windows.ps1" | iexOr download the latest Windows .exe installer from GitHub Releases.
2. Create two OpenAI tunnels
Open:
https://platform.openai.com/settings/organization/tunnels
Create two separate tunnel IDs:
Local Files
ChromeUse a different tunnel ID for Local MCP and Web MCP if both will run at the same time.
3. Create a Runtime API key
Open:
https://platform.openai.com/settings/organization/api-keys
For least privilege, use a restricted key whose principal has:
Tunnels: Read
Tunnels: UseDo not use an OpenAI Admin API key as the long-running tunnel Runtime API key.
When Remember API key securely is enabled, the bridge stores it using Electron safeStorage / the operating system's credential encryption rather than putting the secret directly into normal settings.
4. Configure the desktop app
Enter:
Runtime API key
Local tunnel ID
Web tunnel ID
One or more allowed local folders
Read-only/read-write mode per folder
Whether Local MCP may write files
Whether Local MCP may run terminal commands
Whether Web MCP coordinate/vision tools are enabled
Optional auto-start at login
Then start either or both services:
Local MCP ON
Web MCP ON5. Add the MCP connections in ChatGPT
In ChatGPT, enable Developer mode under Settings → Security and login, then open Plugins and create MCP connections using Tunnel as the connection method. Availability can depend on account/workspace policy.
Local connection:
Name: Local Files
Connection: Tunnel
Tunnel: <your Local tunnel ID>Browser connection:
Name: Chrome
Connection: Tunnel
Tunnel: <your Web tunnel ID>Start the matching service in ChatGPT MCP Bridge before refreshing/scanning its tools in ChatGPT.
What gets installed
Packaged releases include the runtime pieces needed by the bridge:
Desktop tray app
Local MCP server
Chrome DevTools MCP
Electron/Node runtime
Full OpenAI
tunnel-client
You do not need to manually install Node.js, npm, Go, the MCP SDK, Chrome DevTools MCP, or tunnel-client when using a packaged release.
On supported Linux systems, the one-command installer also installs required system packages and installs Google Chrome Stable automatically if Chrome/Chromium is not already available.
Local MCP
Files and folders
Multiple allowed local roots
Read-only/read-write mode per folder
List files and folders
Inspect file metadata
Read UTF-8 text files
Recursive filename/content search
Create folders
Create or replace text files
Filesystem path traversal/symlink checks
Shell
Terminal execution is optional.
Restricted shell
Designed for common development work while blocking several obvious destructive/system-level commands.
Examples:
npm install
npm run build
git status
git diff
git pull
git push
python script.py
adb devicesRestricted shell is a guardrail, not a complete OS sandbox.
Full shell
Full shell executes Bash on Linux or PowerShell on Windows with the current user's privileges.
Treat Full Shell as local-user-equivalent code execution. Enable it only for a machine and MCP connection you trust.
Web MCP / Chrome
Web MCP uses chrome-devtools-mcp and your existing Chrome session, so ChatGPT can work with tabs and authenticated sessions already open on your computer.
Capabilities include:
navigate/open/close/select tabs;
click, hover, type, fill forms, and send keyboard input;
screenshots and accessibility snapshots;
console and network inspection;
performance/Lighthouse tooling;
optional coordinate/vision tools;
optional usage-statistics opt-out.
Chrome setup
Open:
chrome://inspect/#remote-debuggingEnable remote debugging for your local Chrome profile.
When Coordinate/vision tools is enabled, ChatGPT MCP Bridge launches Chrome DevTools MCP with:
--experimentalVisionDesktop app
Linux and Windows
System tray/taskbar controls
Start/stop Local MCP independently
Start/stop Web MCP independently
Restart running services
Optional launch at OS login
Optional MCP auto-start
Runtime logs
Secure Runtime API-key storage
Bundled full OpenAI
tunnel-clientin packaged releases
Windows installer details
The PowerShell installer can:
install Git if missing;
install Node.js LTS if missing;
install Go if missing;
download ChatGPT MCP Bridge;
install npm dependencies;
build the full OpenAI
tunnel-client.exe;validate the application;
build the Windows installer;
launch the generated installer.
If raw.githubusercontent.com is stale or blocked on your network:
Set-ExecutionPolicy -Scope Process Bypass -Force
$script = Invoke-RestMethod -Headers @{ Accept = "application/vnd.github.raw+json" } -Uri "https://api.github.com/repos/haseebgb92/ChatGPT-MCp/contents/install-windows.ps1?ref=main"
Invoke-Expression $scriptBuild from source
Source builds are mainly for contributors. Packaged releases are easier for normal use.
Linux
git clone https://github.com/haseebgb92/ChatGPT-MCp.git
cd ChatGPT-MCp
npm install
npm startWindows
git clone https://github.com/haseebgb92/ChatGPT-MCp.git
cd ChatGPT-MCp
npm install
npm startA source/development run expects a full tunnel-client either on PATH or selected in the app's Advanced settings.
Building installers
GitHub Actions is configured in:
.github/workflows/build.ymlVersion tags build and publish:
Platform | Artifacts |
Linux |
|
Windows | NSIS |
The workflow also compiles and bundles the full OpenAI tunnel-client.
Example release:
git tag v0.1.0
git push origin v0.1.0Local development
npm install
npm run checkLinux:
npm run dist:linuxWindows:
npm run dist:winBuild output is written to dist/.
Security
This project deliberately exposes powerful local capabilities, so use the smallest access level that fits the task.
Never commit Runtime API keys.
Prefer a restricted Runtime API key.
Use separate tunnel IDs for Local and Web MCP.
Prefer read-only roots unless write access is required.
Enable shell only when needed.
Full shell is not a sandbox.
Chrome MCP can act inside logged-in browser sessions.
Review consequential actions before allowing them.
The app redacts OpenAI-style
sk-keys from runtime logs.
See SECURITY.md for details.
Project structure
.
├── .github/workflows/build.yml
├── assets/
├── bundled/
│ └── local-mcp/
├── install-linux.sh
├── src/
│ ├── main.js
│ ├── preload.js
│ └── renderer/
├── .gitignore
├── LICENSE
├── package.json
└── README.mdCurrent status
v0.1.3
The current release focuses on making Local MCP and Chrome MCP simple to run from a tray app on Linux and Windows.
License
MIT License.
Credits
OpenAI Secure MCP Tunnel /
tunnel-clientModel Context Protocol
Chrome DevTools MCP
Electron
Built by Advertpreneur.
This server cannot be deployed
Maintenance
Related MCP Connectors
Use your Mac, Windows or Linux computer from ChatGPT, Claude or Codex: files, commands, documents.
Safe folder access for ChatGPT and Claude: read, write and search files, risky tools opt-in.
Stealth web browser for agents: search, fetch, click, download and type in persistent MCP sessions.
MCP connector that lets ChatGPT list, search, and run your Apple Shortcuts via a local Mac agent
Related MCP Servers
- AlicenseAqualityBmaintenanceEnables ChatGPT web to use local tools like file reading, command execution, and patch application through an MCP server over OpenAI Secure MCP Tunnel.62MIT
- AlicenseAqualityCmaintenanceEnables ChatGPT web to interact with local Windows/WSL shell and code workspaces via an MCP server, providing file access, shell execution, and snapshot-based workspace management with per-command authorization.18MIT
- FlicenseNot gradedqualityBmaintenanceEnables ChatGPT Web to securely access local files and run commands via MCP, with optional OpenCode agent mode for autonomous tasks.-
- FlicenseBqualityAmaintenanceEnables ChatGPT Web to securely control a trusted local computer through an OpenAI Secure MCP Tunnel, letting it perform file, process, Git, and other system operations on Windows, macOS, and Linux.62-