Language-Specific Notes

Python

Manual Connect / Disconnect

import asyncio
from db import Nautilus

async def main():
    client = Nautilus()
    await client.connect()
    try:
        users = await client.user.find_many()
    finally:
        await client.disconnect()

asyncio.run(main())

Auto-Register

If you enable auto-register mode, generated model classes expose a direct Nautilus accessor:

from db import Nautilus
from db.models import User

async with Nautilus(auto_register=True) as client:
    admins = await User.nautilus.find_many(where={"role": "ADMIN"})

recursive_type_depth

This setting is currently Python-only. It controls how many levels of recursive include type helpers Nautilus generates.

Python version baseline

The current generated Python templates import typing.NotRequired directly from the standard library. In practice, Python 3.11+ is the safest baseline for generated clients even though the CLI package itself publishes for older Python versions.

chunk_size on find_many

The current generated Python client exposes chunk_size on find_many and forwards it to the engine as protocol-level chunkSize.

Sync vs Async

If you use interface = "sync", the client surface becomes synchronous instead of async.

Local install behavior

nautilus generate can also perform the target-specific local install step, but the normal workflow is still to import from the configured output directory directly.

Prefer:

nautilus generate --schema schema.nautilus --no-install

The default install step copies the generated package into site-packages/nautilus. That is convenient for some local experiments, but it also means the generated client can shadow the separate CLI shim installed by nautilus python install.

JavaScript / TypeScript

TypeScript declarations

Nautilus writes .d.ts declarations so the generated package is usable in both plain JS and TS projects.

chunkSize on findMany

The current generated JS client exposes chunkSize on findMany and forwards it to the engine at the protocol level.

Select and include

The generated types reflect logical field names from your schema, including mapped fields. That means select and include shapes are schema-aware rather than raw database-column-driven.

Local install behavior

The current generation story is local-first. Import from the configured output directory. If you want generation without target-specific local install behavior, use:

nautilus generate --schema schema.nautilus --no-install

When you keep the default install step, Nautilus can also copy the generated files into the nearest node_modules/nautilus, but output/index.js is still the stable entry point documented here.

Rust

• the generated Rust API is schema-aware rather than raw-SQL-first
• Rust generation can emit bare sources for embedding into an existing workspace
• --no-install skips the target-specific local install behavior if you only want files written to disk

If you want the low-level runtime building blocks instead of generated sources, the upstream workspace also exposes dedicated crates, but that is outside the main end-user scope of this site.

Java

Plain Java Compile / Run

Windows

javac --release 21 -cp "db\dist\nautilus-client.jar;db\dist\lib\*" Main.java
java -cp ".;db\dist\nautilus-client.jar;db\dist\lib\*" Main

POSIX

javac --release 21 -cp "db/dist/nautilus-client.jar:db/dist/lib/*" Main.java
java -cp ".:db/dist/nautilus-client.jar:db/dist/lib/*" Main

• the generated Java runtime loads dotenv data before spawning the engine
• install = true is currently ignored for Java in the upstream runtime notes
• mode = "jar" is the best fit when you want the most explicit local output story