slew docs/start/overview

slew documentation

slew is an EU-native deploy platform: git in, live URL out, on European infrastructure. It is four products behind one account and one CLI:

Product What it does
Deploy Static sites and server runtimes (Node, Next.js) at https://<project>.slew.cloud. Connect a repo for git push-to-deploy with PR previews, or ship a built directory with slew deploy. See how sites are served.
Analytics Cookieless traffic and audience stats for every deployed site, measured at the CDN edge — no script, no cookies, no consent banner.
Share Temporary shares: slew share plan.html puts a throwaway page on an expiring URL — built for coding agents, with an agent skill and a Claude connector.
AI Gateway An OpenAI-compatible endpoint backed exclusively by EU-hosted models.

Sovereignty is the point. All platform data, uploaded artifacts, inference, and traffic stay in the EU — the platform runs on Hetzner (Germany), storage is Hetzner Object Storage, DNS and CDN are Bunny (Slovenia), and every AI model is served by a European provider (Mistral, Scaleway, OVHcloud, IONOS). Identity is not outsourced to US-hosted auth providers: login is the GitHub device flow plus slew's own token system, and tokens are stored only as SHA-256 hashes.

Quickstart

npm install -g slew

slew login              # GitHub device flow — no password, no browser redirect dance
cd my-site
npm run build           # build with whatever you use; slew ships the output
slew init               # create a project named after this directory
slew deploy dist        # tar, upload, live

Deploying 42 files (1.3 MB) to 'my-site'…
Deployed dep_9f3
Live at https://my-site.slew.cloud

That's the whole loop. Every deploy is kept immutably, so rolling back is instant:

slew deployments        # ● marks the live one
slew rollback dep_8a1   # any earlier deployment, live again in seconds

For the AI endpoint, mint a key and point any OpenAI SDK at it:

slew token ai "local dev"

from openai import OpenAI
client = OpenAI(base_url="https://ai.slew.cloud/v1", api_key="slew_...")
resp = client.chat.completions.create(
    model="slew/mistral-small",
    messages=[{"role": "user", "content": "Hallo!"}],
)

How it fits together

slew is four hostnames:

Hostname What it is
api.slew.cloud The control plane the CLI and dashboard talk to: projects, deployments, tokens, usage. See the HTTP API reference.
*.slew.cloud Every project gets a subdomain, served from immutable per-deployment file trees with CDN-friendly caching. See How sites are served.
ai.slew.cloud The OpenAI wire format in front of EU inference only. See AI gateway.
app.slew.cloud The web console: projects, deployments with one-click rollback, traffic and audience analytics, AI usage.

Deploys flow one way: the CLI uploads your built site as a tarball to the API, the API extracts it into EU object storage as an immutable file tree, and the router serves whichever tree is currently active — which is why rollback is instant and old deployments never change.

What slew is (and isn't) today

Currently in the platform:

  • Static site deploys — build on your machine or in CI, slew deploy ships the output. Instant rollback to any previous deployment.
  • Server runtimes — Node and Next.js apps run on slew's own sandboxed runner nodes, with env vars and runtime logs in the console.
  • Git push-to-deploy — connect a GitHub repository; every push builds and deploys on the platform, pull requests get expiring preview URLs.
  • Temporary shares — one-shot preview deploys on a throwaway tmp-*.slew.cloud URL that expires after 1–30 days.
  • Custom domains — attach your own domain to a project with free TLS.
  • Analytics — cookieless traffic and audience stats on every project: requests, bandwidth, pageviews, unique visitors, top pages, referrers, and countries, in the console's Traffic tab, slew stats, and the API.
  • AI gateway — eight slew/ models across four EU providers, metered per account, with a per-account rate limit and daily token budget.
  • Organizations — shared team workspaces with roles and email invites, on flat plans with no seat fees. See Organizations.
  • CLI, console, and HTTP API — everything above without hand-rolling REST calls.

Privacy commitments

  • The AI gateway never logs prompt or completion contents — usage records contain token counts and model names only.
  • Analytics are cookieless and script-free: nothing runs in your visitors' browsers, no IPs or user agents are stored, and unique visitors are counted via transient daily-salted hashes whose salts are destroyed within days.
  • Token secrets are shown once and stored only as SHA-256 hashes.
  • Deployment uploads are extracted with strict limits and path sanitization, stored per deployment, and never mutated.