Deployment

One Next.js app serves everything: the public site, the /admin console, the REST API, and the MCP server. One Postgres database holds all content. There are no required third-party services. Pick whichever path fits your infrastructure.

Path A — VPS with Docker Compose (recommended)

Any 2-core VPS works. You need a domain with an A record pointing at the server. The repo ships docker-compose.yml (app + Postgres) and a Caddyfile that provisions HTTPS automatically.

git clone <your-repo> && cd <your-repo>
cp .env.example .env
# edit .env: DOMAIN, POSTGRES_PASSWORD, BETTER_AUTH_SECRET (openssl rand -base64 32)
docker compose up -d

The first boot builds the app against the live database (a few minutes) and pushes the schema. Optionally load the starter content:

docker compose exec app pnpm --filter @repo/database seed

Then open https://your-domain/admin — the setup wizard creates the owner account and saves your site identity.

Path B — Vercel + Neon

1. Create a Neon Postgres database; copy the connection string.
2. Import the repo into Vercel with root directory apps/web.
3. Set the environment variables from .env.example (at least DATABASE_URL, NEXT_PUBLIC_SITE_URL, BETTER_AUTH_SECRET, BETTER_AUTH_URL).
4. Push the schema once from your machine:

DATABASE_URL=<neon-url> npx prisma db push --schema=packages/database/prisma/schema.prisma
DATABASE_URL=<neon-url> pnpm --filter @repo/database seed   # optional starter content

5. Deploy, then visit /admin for the setup wizard.

Path C — bare VPS with PM2

For a plain Node server with no Docker. Install Node 20+ and pnpm, provision Postgres, then build and run under a process manager.

pnpm install
pnpm --filter @repo/database build          # prisma generate
npx prisma db push --schema=packages/database/prisma/schema.prisma
pnpm --filter @repo/database seed            # optional
pnpm --filter web build                      # content-collections + next build

Run the built app with PM2. The start script defaults to port 3000, so pass -p 3001 (or your chosen port) explicitly and put a reverse proxy in front for TLS:

// ecosystem.config.js
module.exports = {
  apps: [
    {
      name: "directory",
      cwd: "./apps/web",
      script: "node_modules/.bin/next",
      args: "start -p 3001",
      env: {
        NODE_ENV: "production",
        DATABASE_URL: process.env.DATABASE_URL,
        NEXT_PUBLIC_SITE_URL: process.env.NEXT_PUBLIC_SITE_URL,
        BETTER_AUTH_SECRET: process.env.BETTER_AUTH_SECRET,
        BETTER_AUTH_URL: process.env.BETTER_AUTH_URL,
      },
    },
  ],
};

pm2 start ecosystem.config.js

Reference env vars from the process environment rather than hardcoding them in the config file. See Environment variables for the full list.

Scheduling the pipelines

The pipelines are POST /api/cron/* endpoints. Nothing calls them for you — you schedule them with server crontab, GitHub Actions, or an agent. Each call needs an API key with the publish scope (create one in /admin/api-keys).

curl -X POST https://your-domain/api/cron/news \
  -H "Authorization: Bearer $API_KEY"

A sensible starting schedule:

Pipeline	Endpoint	Suggested cadence
News	/api/cron/news	Hourly
News enrich	/api/cron/news-enrich	Every 2 hours
Tools crawl	/api/cron/tools-crawl	Daily
Tools enrich	/api/cron/tools-enrich	Daily
Data audit	/api/cron/data-audit	Nightly
Tool discovery	/api/cron/tool-discovery	Daily
Compare generate	/api/cron/compare-generate	Weekly
Stacks refresh	/api/cron/stacks-refresh	Weekly
Roles sync	/api/cron/roles-sync	Weekly
Tags sync	/api/cron/tags-sync	Daily

Example crontab lines:

0 * * * *   curl -sS -X POST https://your-domain/api/cron/news          -H "Authorization: Bearer $API_KEY"
0 */2 * * * curl -sS -X POST https://your-domain/api/cron/news-enrich    -H "Authorization: Bearer $API_KEY"
30 3 * * *  curl -sS -X POST https://your-domain/api/cron/tools-crawl    -H "Authorization: Bearer $API_KEY"

Every run is recorded at /admin/pipelines with status, duration, and item counts.