Design the database. Ship the backend.
Model your schema visually, put every change — human or AI — through review, generate a production Go backend you own, and deploy it to your own server with one command.
One model drives the whole backend.
Model it
Design entities, relationships, and nested structures in a visual editor — or import an existing MySQL or PostgreSQL database to start.
Review every change
Schema and data changes — from teammates or AI agents — land as change requests: structured diffs you approve before anything is applied.
Generate the backend
Deterministic templates turn the approved model into production Go — migrations, gRPC + REST API, auth, Docker, CI. Same model in, same code out.
Deploy it live
One command provisions a server, applies the schema, and puts your API live behind HTTPS — on infrastructure you own.
Everything between an idea and a running backend.
Six pieces, one model. Use them together or pick what you need.
Visual modeling
Entities, relationships, nested JSON structures, and extended field types — versioned like code.
Explore the editor →Change requests
Pull requests for your database. Every schema and data change is a reviewable diff, with a full history of who proposed it and who approved it.
See the workflow →Go code generation
A complete backend from your model: type-safe sqlc queries, gRPC + REST, migrations, auth, caching — powered by an open-source engine.
See what's generated →One-command deploy
From approved schema to a live HTTPS API on your own server, or a VM created with your own cloud account. nuzur never holds your credentials.
How deploy works →Automations & webhooks
Fire HMAC-signed webhooks when approved changes are applied — with retries, delivery logs, and secrets you control.
Read the docs →Data manager
Query, edit, and validate data in a visual grid — even on private databases, through an outbound-only local agent.
Explore features →Let agents build. Keep the approval.
Connect Claude Code, Cursor, VS Code, or any MCP client to nuzur's hosted MCP server. Agents can design schemas, stage data changes, and even drive a full deploy — but every change waits for your review.
The agent never holds the pen
Agent credentials can propose, not execute. Nothing touches your database without a human approving it.
Granular MCP tools
Purpose-built tools for schema edits, data change requests, queries, automations, and deployments.
Deploys included
A built-in MCP prompt walks your agent through an entire production deploy — with you at every gate.
From approved model to live API in one command
nuzur-cli deploy provisions the server, installs the database locked to localhost, builds and runs your generated API behind auto-HTTPS, and pairs it all back to nuzur through an outbound-only agent — no inbound database ports. The generated source lives in a directory you own, and re-deploys preserve your custom code.
Your own Linux server, or a VM created through your own CLI on DigitalOcean, Hetzner, Linode, Google Cloud, Azure, Vultr, or Scaleway.
Prefer to talk? Claude can drive the whole deploy through the nuzur MCP server.
$ nuzur-cli deploy
→ generating app (go-code-gen)
→ provisioning server
→ database up · localhost only
→ building image · starting api
→ applying schema · pairing agent
✓ live at https://api.yourapp.dev
Code you own, not a backend you rent
nuzur generates a real codebase in your repository. There's no runtime dependency on nuzur — you're never tied to the platform.
A rented backend
- Runtime lives on their infrastructure
- API surface defined by the platform
- Pricing scales with their metering
- Leaving means rebuilding
nuzur
- Plain Go + sqlc in your repository — no ORM, no reflection
- Deterministic templates — same model in, same code out, never AI-generated
- A standard Dockerfile — deploy it anywhere Docker runs
- Stop using nuzur tomorrow and everything keeps running
Reviews, roles, and history built in
Share projects with your team, control who can edit and who can approve, and read your schema history like a git log.
Start with the model. Everything else follows.
Free for personal use — modeling, MCP server, code generation, and deploy included.
Start freeMySQL & PostgreSQL · the code is yours