Quickstart
The shortest path is GitHub npx, initialize a local policy, run the HTTP MCP server, then create a ChatGPT Custom Connector.
Option A: temporary GitHub npx install
No clone is required for this path.
npx github:Harzva/chatgpt2localbridge init --root ~/Projects
set -a; source .env.local; set +a
npx github:Harzva/chatgpt2localbridge --http 3838
Option B: clone and run
git clone https://github.com/harzva/chatgpt2localbridge.git
cd chatgpt2localbridge
npm install
npm run build
node dist/index.js init --root ~/Projects
set -a; source .env.local; set +a
node dist/index.js --http 3838
Health check
curl -sS http://127.0.0.1:3838/health
Authentication choice
Choose OAuth for any public HTTPS tunnel. Choose No Authentication only for loopback or short-lived private tests where the endpoint is not publicly reachable.
Public tunnel: OAuth
Loopback-only test: No Authentication
Advanced split policy: Mixed Authentication
Linux server
Run a separate bridge and connector on Linux when ChatGPT needs Linux paths such as /srv/workspace or /home/agent/projects.
REMOTE=linux-box \
REMOTE_WORKSPACE=/srv/workspace \
REMOTE_ALLOWED_ROOTS="/srv/workspace,/home/agent/projects" \
PUBLIC_BASE_URL=https://linux-bridge.example.com \
bash scripts/deploy-linux-bridge.sh
Local console
Open the local operator console to inspect status, tool calls, file-download events, and audit history.
http://127.0.0.1:3838/app
Native Policy Center
Install the macOS app when you want a local desktop console for policy editing, trace export, project bundles, cloud downloads, and the MCP tool catalog.
npm run macos:install
open /Applications/ChatGPT2LocalBridge.app
Skill root
The app and policy support a dedicated local skill root. Prefer this path, not the whole Codex runtime directory.
{
"skillRoots": [
"/Users/YOUR_USERNAME/.codex/skills"
]
}