Linux Deployment

Run the bridge on the Linux machine that owns the files, expose only the MCP endpoint through HTTPS, and create a separate ChatGPT connector for Linux.

Route

ChatGPT

Linux Connector

HTTPS Tunnel

localhost:3838

Linux MCP

Approved Roots

Deploy

REMOTE=linux-box \
REMOTE_WORKSPACE=/srv/workspace \
REMOTE_ALLOWED_ROOTS="/srv/workspace,/home/agent/projects" \
REMOTE_RUNTIME=/data/chatgpt2localbridge-runtime \
PUBLIC_BASE_URL=https://linux-bridge.example.com \
bash scripts/deploy-linux-bridge.sh

Expose HTTPS

cloudflared tunnel --url http://127.0.0.1:3838 --no-autoupdate

For long-running use, prefer a named Cloudflare Tunnel, fixed ngrok domain, or your own reverse proxy.

Create Connector

Name: ChatGPT2LocalBridge Linux
Server URL: https://linux-bridge.example.com/mcp
Authentication: OAuth

Test

Use the ChatGPT2LocalBridge Linux connector. Call file.read_path, not cloud Python.

paths=["/srv/workspace/README.md"]
maxLines=40