Qwen Code

Skip it — one-line setup

The script installs Qwen Code from the official npm source, writes the gateway settings into the tool's own config file (our settings: ~/.qwen/settings.json plus ~/.qwen/.env, backing up any existing ones), installs a qwen-shinloo launcher that pins traffic to us, and runs a live check. It asks for your gateway key, or take it from --key / --key-file. --uninstall restores exactly what it wrote.

curl -fsSL https://quantoport.com/agents/qwen-code/setup | bash

1 · Install

Install from the official npm source (this package requires Node.js 22 or newer).

node -v                          # needs v22 or newer
npm install -g @qwen-code/qwen-code
qwen --version

2 · Configure — point it at QuantoPort

Gateway base URL https://quantoport.com/v1; get your key from the console.

1) Write ~/.qwen/settings.json (selects the protocol, endpoint and default model; your other provider entries are kept):

{
  "modelProviders": { "openai": [ { "id": "qwen3.8-max", "baseUrl": "https://quantoport.com/v1", "envKey": "OPENAI_API_KEY" } ] },
  "security": { "auth": { "selectedType": "openai" } },
  "model": { "name": "qwen3.8-max" }
}

2) Write ~/.qwen/.env for the key (create it if missing; the block this script writes is marked managed and is the only part removed by --uninstall):

OPENAI_API_KEY="YOUR_API_KEY"
OPENAI_BASE_URL="https://quantoport.com/v1"
OPENAI_MODEL="qwen3.8-max"

On the China endpoint use https://api.shinlinkai.com/v1 instead.
3) A qwen-shinloo launcher is installed in ~/.local/bin/: it exports this key and endpoint before starting qwen.
4) Reopen the terminal and run qwen (or qwen-shinloo); switch models in-session with /model.

3 · Verify

One request proves you are on our channel.

qwen-shinloo -p "reply with OK"       # recommended: pins our channel
qwen -p "reply with OK"                 # also works once the config is written

4 · Common pitfalls

• You need Node.js 22+: the package pins engines >=22 and will refuse to install on older Node. Check with node -v first.
• OPENAI_BASE_URL must end with /v1; the bare gateway root returns 404.
• settings.json and .env are read at startup — reopen the terminal after editing them.
• If OPENAI_API_KEY is already exported in your environment (or the tool falls back to its own default channel), that wins over the config files and the request will go elsewhere — launch qwen-shinloo instead to pin our channel.
• If the file already had OPENAI_* lines, the script comments the old ones out and notes the original value, so nothing is silently overwritten.
• Model names follow GET /v1/models for your key — the published catalogue can change with upstream supply.