What LuisSystemd is actually for
Not a toy. A real multi-tenant hosting platform: an account gets a persistent workspace with a real always-on server, a real SQLite database, a real terminal, and a real public URL — without ever touching a VPS, a Dockerfile, or a systemd unit file.
Use it when...
You're building something that calls an LLM
Encrypted secrets with real precedence rules, preset keys for OpenAI/DeepSeek/Anthropic/Stripe, and an AI assistant that writes code referencing your secrets by name — never the value. Deploy a real endpoint that calls a real model in under a minute.
You want a public URL for a demo, now
lsd init --template flask-api && lsd deploy and you have a live, shareable URL. No cloud console, no billing setup, no waiting on a build pipeline.
Your project needs a real database, not a spreadsheet
Any .db file in your workspace is a real SQLite database, browsable in the Files tab — tables, row counts, ad-hoc SQL — and readable from your deployed server's own code at the same path.
Something needs to run on a schedule
Backups, scrapes, cleanup jobs — lsd jobs add "python3 backup.py" --every 60 and a background scheduler runs it for you, tracking exit codes and output, no cron access required.
You're prototyping and don't know the shape yet
Start from a template (Flask, Express, a bare Cloudflare-Workers-style worker.js, or static HTML) or blank. Swap approaches without provisioning anything new — it's still the same project, same workspace.
More than one person needs access
Real per-project collaborators via invite links, not shared credentials. Multi-tenant by design — your projects, files, and secrets are scoped to your account, invisible to everyone else.
How it comparesPicking the right tool for the job
| If you need... | Reach for |
|---|---|
| Static hosting live in seconds, same account as everything else | LuisSystemd Share — no separate host, no git repo, just files you already have in your workspace. |
| Serverless functions at real production scale | Cloudflare Workers / actual serverless platforms — the worker.js template here is for local dev convenience, the runtime is a Node shim on shared infra, not the real edge. |
| A quick backend + database + terminal, own account, minimal setup | LuisSystemd — this is exactly the gap it fills. |
| Scheduled jobs without configuring cron/systemd yourself | LuisSystemd Scheduled Jobs. |
| A public URL without babysitting your own tunnel process | LuisSystemd manages bore for you and re-tunnels automatically on redeploy — no manually re-copying a fresh URL every time a quick tunnel expires (a real, repeatedly-hit annoyance with e.g. bare cloudflared --url quick tunnels, which silently die on restart and take down anything hardcoded to point at them). |
| Guaranteed uptime SLAs, dedicated resources, isolation | Not this — see below. |
What it's not for
- Production workloads with uptime guarantees. It runs on shared infrastructure (one VM, all tenants) — there's no SLA, no dedicated resources, no isolation beyond per-account file/db/secret scoping.
- High-traffic public services. Fine for demos, internal tools, and small real projects; not sized for anything that needs to scale horizontally.
- Anything that needs guaranteed persistent compute performance.
lsd statusshows real, honest instance-wide stats (CPU/mem/disk are shared, labeled as such) — noisy-neighbor effects are possible. - Long-term secret storage for production credentials. Secrets are genuinely encrypted at rest, but this is a dev/prototyping platform first — treat production-critical credentials with the same caution you would on any shared host.
Why bore for tunneling?
Every deployed server and every static share gets its public URL from bore, a small open-source TCP tunnel — not a custom proxy layer. It's a deliberate choice: bore is a single static binary, needs no account or API key of its own to open a tunnel, and hands back a plain host:port the moment a process starts listening — which is exactly the shape LuisSystemd needs to go from "your code just started" to "here's a real public URL" with nothing in between.
The honest pitch
It's the platform for the gap between "just run it on my own machine" and "set up real cloud infrastructure for a thing that doesn't need it yet." A real account, a real persistent workspace, a real deployed process with a real public URL — reachable from a one-line install and three CLI commands, with encrypted secrets and a database along for the ride.