EngineView

A multi-user dashboard for Cloudflare Analytics Engine. Cloudflare gives you a powerful time-series store but no UI and no way to save queries. EngineView lets people sign up, connect their own Cloudflare account, run SQL against their datasets, save the queries they care about, and chart them on multi-panel dashboards.

Each user brings their own Cloudflare account. API tokens are encrypted at rest and only ever decrypted on the server to run a query, so a token is never sent to the browser or shared between users.

Screenshots

Write SQL against your datasets and switch between a table and a chart:

Arrange saved queries as stat tiles and charts on a dashboard with a shared time range:

Status

Multi-user accounts, per-user Cloudflare connections, the SQL query runner with charts and stat tiles, and multi-panel dashboards are all in place.

Stack

• Nuxt 4 (Vue 3) for the app and server API
• better-auth for email and password accounts
• Postgres via Drizzle ORM
• Resend for verification and password-reset email

How it works

• Users sign up and log in (better-auth, signed session cookies). Email addresses are verified before first sign in, and passwords can be reset, with the emails sent through Resend.
• Each user adds one or more Cloudflare connections (account id + API token). The token is encrypted with AES-256-GCM before it is stored.
• A query is sent to the server, which decrypts the relevant token, calls the Analytics Engine SQL API, and returns the rows. The token never leaves the server.
• Saved queries and dashboards are stored in Postgres, scoped to each user.
• A result can be charted (line, area or bar) or shown as a single stat tile. Saved queries become dashboard panels, laid out in a grid with a shared time range. A query can use the $SINCE token (e.g. timestamp > $SINCE), which expands to the dashboard's selected range.

Setup

Prerequisites: Node 20+, a Postgres database, and a Cloudflare account using Analytics Engine.

npm install

# 1. Configure the environment
cp .env.example .env
#    Set DATABASE_URL, BETTER_AUTH_SECRET, BETTER_AUTH_URL and ENCRYPTION_KEY.
#    Generate secrets with: openssl rand -hex 32

# 2. Apply the database schema
npm run db:migrate

# 3. Run locally
npm run dev

A local Postgres is easy to start with Docker:

docker run -d --name engineview-pg -e POSTGRES_PASSWORD=postgres \
  -e POSTGRES_DB=engineview -p 5432:5432 postgres:16

On macOS, if npm run dev fails with a vite-node socket error (EINVAL), your $TMPDIR path is too long for a unix socket. Run it with a short temp dir: TMPDIR=/tmp npm run dev.

Scripts

• npm run dev / npm run build / npm run preview
• npm run db:migrate (apply migrations), npm run db:generate (after a schema change)
• npm run lint, npm run format, npm run typecheck, npm test

Deploy

Yes, it is a Cloudflare tool that does not run on Cloudflare, and that is on purpose. EngineView reads your data through the Analytics Engine SQL API over plain HTTPS, so the edge is never in the request path. That makes it a boringly portable server app: run it on a small VPS, with Docker, or on a host like Fly or Railway. You bring the Cloudflare account; the dashboard lives wherever you want it.

Docker Compose

The simplest way to run the whole stack (Postgres, migrations, app):

export BETTER_AUTH_SECRET=$(openssl rand -hex 32)
export ENCRYPTION_KEY=$(openssl rand -hex 32)
docker compose up --build

The app is served on http://localhost:3000. A migrate step runs before the app starts, and GET /api/health reports database connectivity.

Manual

npm ci
npm run build
npm run db:migrate          # against your production DATABASE_URL
node .output/server/index.mjs

Set DATABASE_URL, BETTER_AUTH_SECRET, BETTER_AUTH_URL and ENCRYPTION_KEY in the environment. The server validates them on startup and exits with a clear message if any are missing. Run it behind a reverse proxy that terminates TLS and forwards the client IP (x-forwarded-for or cf-connecting-ip) so rate limiting works.

License

MIT. See LICENSE.