@cartanova/qgrid-cli

English · 한국어

Run the Qgrid server in one line. An LLM proxy that exposes OpenAI/Anthropic subscription credits as an HTTP API + dashboard.

Install

npm i -g @cartanova/qgrid-cli

Installing also syncs the qgrid skills automatically:

• global install: ~/.codex/skills/qgrid, ~/.claude/skills/qgrid
• project install: .agents/skills/qgrid, .claude/skills/qgrid

Project installs create symlinks, falling back to copies when symlinking fails.

Preparing PostgreSQL

Qgrid needs PostgreSQL to store OAuth tokens and request logs. If you already have a reachable PostgreSQL, connect with --db or the QGRID_DB_* environment variables. If you don't have one locally, you can spin one up with Docker:

docker run --name qgrid-postgres \
  -e POSTGRES_USER=postgres \
  -e POSTGRES_PASSWORD=postgres \
  -e POSTGRES_DB=qgrid \
  -p 5432:5432 \
  -d postgres:18

Usage

# Run with a DB URL
qgrid --db postgres://user:password@host:port/dbname

# Specify a port
qgrid --db postgres://... -p 3000

# Configure the DB with environment variables (flags can be omitted)
export QGRID_DB_HOST=dev.example.com
export QGRID_DB_PORT=5432
export QGRID_DB_USER=postgres
export QGRID_DB_PASSWORD=postgres
export QGRID_DB_NAME=qgrid
qgrid

The CLI translates the public QGRID_DB_* settings into Sonamu's internal SONAMU_DB_* variables. The packaged CLI defaults to NODE_ENV=production; set NODE_ENV=staging only when intentionally running it as a remote non-production service. This profile selection never creates or renames the database selected by QGRID_DB_NAME. The packaged CLI rejects development and test, which require the source development layout.

Once the server is up, open the dashboard at http://localhost:44900 → register tokens through OAuth login. With -p, --port set, connect on that port instead.

The server port is determined only by the default 44900 or --port. The PORT environment variable is not read as CLI server port input. If the selected port is already in use, the CLI exits with an error instead of killing the existing process or falling back to another port.

Stop with Ctrl+C.

Options

qgrid [options]

  --db <url>         PostgreSQL connection URL
  -p, --port <port>  Server port (default: 44900)
  --skip-update      Skip the automatic update check
  -V, --version      Print version
  -h, --help         Help

On startup the CLI checks the latest version on npm and self-updates whenever the installed version differs, including patch releases. The updater installs the exact published version it just resolved and restarts from the updated binary.

Environment variables

Without the --db flag, DB connection info is read from:

Server behavior variables:

Prerequisites

• Node.js >= 22
• PostgreSQL
• Docker (if running PostgreSQL locally as a container)
• Claude Code (for Anthropic models)

How it works

The CLI ships the Sonamu-based server as a built-in bundle. On launch:

1. Verify the DB connection
2. Start the server (API + dashboard web UI)
3. Load registered tokens from the DB. Later token additions/changes propagate to the running server in real time via PostgreSQL LISTEN/NOTIFY
4. OpenAI tokens: call https://chatgpt.com/backend-api/codex/responses directly over HTTPS and decode SSE. Token-level concurrent permits use smooth weighted routing; a 50-item queue waits at most 60 seconds. Full history is replayed with opaque sessionKey-derived prompt-cache affinity, and AbortSignal cancels queued or active work
5. Anthropic tokens: spawn a fresh, isolated claude process per request (stream-json in/out). Tokens are selected least-used-first. OAuth tokens are refreshed automatically
6. Tokens over their quota threshold (default 80%) are excluded from routing (usage-lookup failures fail open)

The Qgrid app itself does not depend on Docker, but PostgreSQL is required. If you don't run PostgreSQL locally, running it in Docker is the simplest setup.

OpenAI login uses direct PKCE OAuth, and quota checks call the private wham/usage endpoint directly with Codex CLI identity headers. These ChatGPT backend interfaces are undocumented and may change without notice. Mocked tests do not establish live-provider compatibility.

SDK integration

Once the server is running, call it with @cartanova/qgrid-ai-sdk:

import { generateText } from "ai";
import { qgrid } from "@cartanova/qgrid-ai-sdk";

const { text } = await generateText({
  model: qgrid("openai/gpt-6-luna"),
  prompt: "What's the weather in Seoul?",
});

See the @cartanova/qgrid-ai-sdk README for details.