We've been building a cloud controller for OpenWrt and it's now open for signups.
**What it is**
A TR-369/USP agent for OpenWrt that talks to a hosted controller over WebSocket
with mutual TLS. Device identity is a real per-device certificate — the agent
generates its own keypair on first boot, enrols over EST (RFC 7030), and the
private key never leaves the router. First enrolment binds the serial, so the
credential shipped in the package can't be replayed to claim someone else's
device.
On top of that:
- device + client inventory (OUI / DHCP / mDNS fingerprinting)
- application visibility via Aether / Parental Control
- per-device and per-profile policy
- WiFi config, mesh 80211S, also prplos support for (EasyMesh / IEEE 1905), multi-AP
- and the rest of the protocol zoo when you need it — NETCONF, SNMP, gNMI, CoAP/LwM2M
**Building it**
Everything device-side is on GitHub. Standard package feed:
git clone https://github.com/optim-enterprises-bv/optim-wrt.git
Agent: https://github.com/optim-enterprises-bv/ac-client
(the agent is BUSL-1.1 — source is open to read and build, commercial use is licensed.)
Built and tested on BPI-R4 (mediatek/filogic) and IPQ5018.
**Pricing**
$5/mo for up to 3 devices, $10 for 5, $20 for 25. Per-device rates for operators
from $0.50 (100-device minimum). 7-day trial.
https://www.aether-io.com
**Where it's at**
Early but real. The device path — enrolment, mTLS, USP session, DPI reporting —
works and we've run it end to end on actual hardware. We haven't run it at scale
yet, so if you point a pile of routers at it you'll find edges we haven't. That
feedback is exactly what we're here for.
Happy to answer anything.
If we are interested in using an Open Source version of this can we host our own on-prem version of acs.optimcloud.com? As a residential (personal, non-commercial) open-source enthusiast, I don't want management to go outside my own network, but I am interested in managing multiple wi-fi APs running OpenWRT.
There is not to my knowledge currently any ACS/TR181 open source solution, next best would be genieACS but thats TR-69, While OptimACS, like Aether can be self-hosted, they are commercial products. We have not produced an open source version of OptimACS, It is a commercial closed source application. While I understand your wanting to self-host it, unless you have a public ip or other avenue in, your potentially loosing the remote management and other features we provide in Aether itself. ACS does management ok by itself, but not in the way we do it with a full menued interface. We would have to consider the business impact of years of development to release a self-hosted limited version of it and a limited one time binary cost.
I am not intimately familiar with ACS/TR181 but the general outcome appears to follow the management paradigms of OpenSOHO or perhaps OpenWhisp both of which are Open Source. And I would argue that people come to OpenWRT to gain personal control over their own hardware, which also encourages open controllers.
Hosting applications is not the barrier here - it is ceding control to an unknown 3rd party without visibility into the security/maturity of the codebase. Especially in 2026 when many SaaS offerings are springing up based on LLM output, I really want to see a well-maintained, well-planned codebase before providing network access.
I'm not sure how this is relevant to OpenWRT, if it is not an open solution.
Our code base is 8 years old, well before LLM, and AI existed. And OpenWRT was a natural fit for the likes of Carriers, WISPs, Telecomms, as its the base for the likes of QSDK, PRPL, OpenWIFI and many others, OpenWRT runs across the board in many industries, not just the home, where you want to control your own. It exists in many commercial applications also. So adding OpenWRT in was just the side effect for those wanting complete remote control and enhanced capabilities. We wount mention integrating IOT, and Cameras into a single pane of glass that most homes could benefit from. And after years in the telecomm industry, short of openwisp and opensoho, not at scale really exists, commercial USP/ACS servers run in the 10s of thousands of dollars in licensing fees per year. Our primary design was based on one controller for all wireless devices, regardless of being WRT, RDK, Plume, PRPL, QSDK derivitives. If it benefitted the home users at the costs we set then great. and actually the client code originated 19 years ago