aiopynetbox

Fully async NetBox API client for Python, built on httpx2.

Inspired by pynetbox, redesigned for asyncio. This is not a port: pynetbox's core ergonomics (lazy attribute fetches, len() on result sets, sync generators) depend on Python protocols that cannot be awaited, so the API surface here is deliberately different. All I/O is explicit and awaitable, and nothing does network I/O behind your back.

Requirements

• Python 3.11+
• NetBox 3.x / 4.x (v1 and v2 nbt_ API tokens both supported)

Installation

uv add aiopynetbox   # or: pip install aiopynetbox

Quick start

import asyncio
import aiopynetbox


async def main():
    async with aiopynetbox.api("https://netbox.example.com", token="...") as nb:
        # single object
        device = await nb.dcim.devices.get(name="sw-1")
        print(device.name, device.status, device.site)

        # filtered query, pages are fetched concurrently
        async for iface in nb.dcim.interfaces.filter(device_id=device.id):
            print(iface.name)

        # diff-based save: only changed fields are PATCHed
        device.serial = "ABC123"
        await device.save()


asyncio.run(main())

Coming from pynetbox

The traversal (nb.dcim.devices), diff-based save(), and exception taxonomy all carry over. What changes is that implicit I/O becomes explicit:

Nested records come back brief (as NetBox sends them). Touching a field that isn't loaded raises AttributeError telling you to await record.full_details(). It never fires a hidden HTTP request.

Features

• Explicit async everywhere: httpx2.AsyncClient under the hood, used as an async context manager so the connection pool closes deterministically.
• Concurrent pagination: after the first page reveals the count, the remaining pages are fetched in parallel (bounded by max_concurrency, default 4) and yielded in order.
• Cursor pagination (NetBox 4.6+): pass pagination="cursor" to the client to page with the start cursor instead, constant-time per page on very large tables (pages are sequential in this mode, since each cursor comes from the previous response).
• Diff-based writes: save() PATCHes only what you changed, with NetBox's custom_fields merge semantics handled correctly.
• Bulk operations: await nb.dcim.devices.filter(status="offline").update(comments="audit"), await recordset.delete(), and list forms on the endpoint (endpoint.update([...]) / endpoint.delete([...])).
• IPAM allocation: await prefix.available_ips.create() / .list(), plus available_prefixes and available_vlans.
• Plugins: nb.plugins.<plugin>.<endpoint> and await nb.plugins.installed_plugins().
• Choices: await nb.dcim.devices.choices() from OPTIONS metadata.
• Retries with backoff: 429 responses are retried automatically for any method (honoring Retry-After); transient 502/503/504 and connection failures are retried for GETs only, since an ambiguous write may already have been processed. Exponential backoff with jitter; tune with retries=, disable with retries=0.
• Optimistic locking (NetBox 4.6+): records fetched from a detail endpoint remember their ETag; save() sends If-Match, so a concurrent modification fails with a 412 error instead of being silently overwritten. Repeat full_details() calls revalidate with If-None-Match, so unchanged objects aren't re-downloaded or re-parsed.
• Data source sync: await data_source.sync.create().
• Custom models: aiopynetbox.register_model("plugins/bgp", "sessions", BgpSession) maps plugin endpoints to your own Record subclasses; app.endpoint("literal_name") reaches endpoint slugs that contain real underscores.
• Typed: full type hints and a py.typed marker, plus generated hints so IDEs autocomplete endpoint names (nb.dcim.devices) and per-endpoint kwargs (filter(name=...), create(device_type=...)). Hints never restrict anything at runtime: unknown endpoints, lookup expressions, and custom-field filters keep working.

API tour

async with aiopynetbox.api(url, token=token) as nb:
    # read
    device = await nb.dcim.devices.get(123)  # by id (None if missing)
    device = await nb.dcim.devices.get(name="sw-1")  # by filter (ValueError if >1)
    total = await nb.dcim.devices.count(site="main")
    async for d in nb.dcim.devices.filter(status="active", tag=["prod", "core"]):
        ...
    async for d in nb.dcim.devices.all(limit=100, offset=200):  # single page
        ...

    # write
    new = await nb.dcim.devices.create(name="sw-9", device_type=12, site=1, role=3)
    device.serial = "XYZ"
    await device.save()  # PATCH {"serial": "XYZ"}
    await device.update({"serial": "XYZ", "comments": "..."})
    await device.delete()

    # bulk
    await nb.dcim.devices.filter(site="old").update(status="decommissioning")
    await nb.dcim.devices.filter(status="decommissioning").delete()

    # ipam allocation
    prefix = await nb.ipam.prefixes.get(prefix="10.0.0.0/24")
    ip = await prefix.available_ips.create()  # next free IP
    ips = await prefix.available_ips.create([{}, {}])  # next two

    # instance info
    print(await nb.version())  # "4.5"
    print(await nb.status())

Long-lived apps (FastAPI, services)

Create the client once and share it; the async context manager is one-shot, so enter it for the app's lifetime, not per request:

@asynccontextmanager
async def lifespan(app: FastAPI):
    async with aiopynetbox.api(url, token=token) as nb:
        app.state.nb = nb
        yield  # handlers use `await app.state.nb...`; pool closes on shutdown

One shared instance is safe under concurrent requests. See examples/fastapi_app.py for a runnable app.

Custom httpx2 client

Pass your own httpx2.AsyncClient for custom SSL, proxies, event hooks, or MockTransport in tests:

client = httpx2.AsyncClient(verify="/path/to/ca.pem", timeout=60)
async with aiopynetbox.api(url, token=token, client=client) as nb:
    ...

Per httpx2 convention, a client you pass in is yours to close: aclose() and the context manager only close clients the Api created itself, so one client can safely back several Api instances.

Response caching is deliberately not built in: NetBox is a source of truth, and the library can't know your staleness tolerance. If you want HTTP caching, pass a client using hishel's AsyncCacheTransport and set the policy yourself.

Development

Managed with uv:

uv sync              # install environment
uv run pytest        # tests (in-memory fake NetBox, no network)
uv run ruff check    # lint
uv run ruff format   # format

License

Apache 2.0, see LICENSE.