Self-hosted, Vercel-native AI localization. Click Deploy. Run npx languine. That's it.

What the button does:

1. Forks this repo to your GitHub account.
2. Creates a Vercel project with apps/web as the root.
3. Provisions serverless Postgres via the Vercel Marketplace — DATABASE_URL is auto-injected.
4. Prompts you for LANGUINE_API_KEY (any random string) and an optional AI_MODEL slug.
5. Builds and deploys. The build automatically runs drizzle-kit migrate against your fresh database.

After the first deploy, do these two things:

1. Enable Deployment Protection. In your Vercel project: Settings → Deployment Protection → turn on Vercel Authentication (or Password Protection). This gates the dashboard and /cli/token. Without this, your API key is publicly visible.
2. Open the dashboard (https://<your-deployment>.vercel.app). It shows a status checklist and ready-to-paste CLI commands.

That's it. No additional integrations to configure — Vercel AI Gateway authenticates automatically via the project's OIDC token.

The self-hosted CLI ships under the selfhosted npm dist-tag so it doesn't disturb users still pointed at the legacy hosted backend. Install it with npx languine@selfhosted ... (or pin "languine": "^4" in your package.json). npx languine@latest continues to resolve to the legacy 3.x CLI for the old hosted service.

In any project that needs translations:

npx languine@selfhosted login --url https://languine.your-team.vercel.app
npx languine@selfhosted init
npx languine@selfhosted translate

languine login opens /cli/token in your browser. Because Deployment Protection is on, only authorized owners can see the page — copy the API key, paste it into the CLI prompt.

languine init creates a project on your deployment via tRPC (project.create) and writes the returned projectId into languine.json. You don't need to provision anything from the dashboard.

For non-interactive use (CI, scripts):

export LANGUINE_BASE_URL=https://languine.your-team.vercel.app
export LANGUINE_API_KEY=<the-key-you-set-on-vercel>
npx languine@selfhosted translate

name: Languine
on:
  push:
    branches: [main]

jobs:
  translate:
    runs-on: ubuntu-latest
    permissions:
      contents: write
      pull-requests: write
    steps:
      - uses: actions/checkout@v4
      - uses: languine-ai/languine@v4
        with:
          api-key: ${{ secrets.LANGUINE_API_KEY }}
          base-url: ${{ vars.LANGUINE_BASE_URL }}
          project-id: prj_xxxxxxx
          create-pull-request: 'true'

• Hosting: Vercel (Next.js 16, App Router, Node.js runtime).
• Database: Serverless Postgres from the Vercel Marketplace + Drizzle ORM.
• Background jobs: Vercel Workflows (workflow SDK) — durable, resumable, observable.
• AI: Vercel AI Gateway (@ai-sdk/gateway + AI SDK v6) — single configurable model.
• Auth: Single LANGUINE_API_KEY for the CLI/Action; the dashboard is gated by Vercel Deployment Protection.

flowchart LR
  Dev[Developer] -->|languine| CLI
  GHA[GitHub Action] -->|x-api-key| Web
  CLI -->|x-api-key| Web[apps/web on Vercel]
  Web --> Workflows[Vercel Workflows]
  Workflows --> Gateway[Vercel AI Gateway]
  Gateway --> Models[OpenAI / Anthropic / ...]
  Web --> DB[(Postgres)]
  Workflows --> DB
  Owner[Owner browser] -->|Deployment Protection| Web

git clone https://github.com/languine-ai/languine
cd languine
bun install
cp apps/web/.env.example apps/web/.env  # fill in DATABASE_URL etc.
bun dev

Set AI_GATEWAY_API_KEY in apps/web/.env so the Gateway can authenticate when not running on Vercel.

bun test                         # everything
bun test --filter @languine/web  # just the web app
bun test --filter languine       # just the CLI

MIT © Midday Labs AB