Getting started
Sahabti runs two things for you: managed open-source engines — databases, caches, vector stores, brokers — provisioned in about 75 seconds, and projects, apps built straight from a GitHub repository and served on a live URL. Both land in your workspace's own private network. This guide takes you from zero to something running.
1. Create an account
Sign up at app.sahabti.com — free for developers and the community, no credit card required. Signing up creates a personal workspace, and its private network is created for you the first time you launch something into it.
2. Launch your first service
From the dashboard, open Resources → New and pick an engine — PostgreSQL 17, MySQL 8.4, MariaDB 11, FerretDB, Valkey 8, Memcached, Qdrant, or RabbitMQ — and a size. Sahabti provisions the container, wires it into your workspace's isolated private network, and publishes a public endpoint for it automatically.
The catalog also lists the engines that aren't live yet; those render as Coming soon and can't be selected.
Every instance you launch is carved out of your account's shared quota (vCPU, memory, disk) — no per-service bill to manage. The Free tier's quota fits a couple of lightweight instances; upgrade when you outgrow it.
3. Connect
Once the instance is running, its connection details are on the instance page: host, port, username, password, and a ready-to-copy connection URI. Watch it get there live — the provisioning log streams each step as it happens.
For PostgreSQL and MySQL you don't need a client at all to look around: the instance's Table editor and SQL editor browse rows and run statements straight from the dashboard.
4. Deploy an app
Connect GitHub on the Integrations page, then Projects → New Project and import a repository. Sahabti detects the stack, builds a container, and serves it over HTTPS at a hostname of its own — and redeploys on every push to the branch you tracked.
Next steps
- Instances — sizes, lifecycle, provisioning steps, and metrics
- Connecting — per-engine connection examples and what's encrypted
- Database tools — table editor, SQL editor, and database users
- Projects — build and deploy from GitHub
- Workspaces & roles — bring your team in
- API tokens — drive all of it from a script or CI
- MCP for AI agents — hand the same surface to Claude, Cursor, or your own agent