What do you think if Peplink publishes an InControl Model Context Protocol (MCP) server? If you find it useful, how will you use it?
Do you use LLMs to develop software to manage your devices using InControl? Which model do you use? What difficulties are you facing?
Hi Michael,
I think this would be good, currently writing tools which interact with the incontrol API and the local API, and often feed it with the correct API calls to make things happen. An MCP with a lean set of API instructions would be really good, along with the OAUTH2 flow example included. I have to givie it the example to stop it from adding a redirect and expecting me to paste the token in every time.
An MCP for local API would also be great.
This would be very useful!!
I would think I use LLM to quiry overall status of unstable links which I need to address.
Ideal workflow as I see would be:
I say any broadband WAN which had been down lets say 5 times a day, Agentic AI opens a ticket with ISP, communicate automatically, confirm link is stable after repair ticket is closed and reopen if not fixed.
This will help to further provide higher quality of supporting large landscape of the fixed devices where cheap broadband connections often flap and fail health checks and it creates a lot of noise, to be able to see these signals will be very useful. Alerts fatigue vs useful alerts vs noise. Build more useful dashboard with little more data not only Online/Offline but unstable sites which needs to be addressed and may be its a ISP change or they will fix it - it can scale now and save money with ISPs.
Hello Michael,
Yes we do - happy to share what our product can do so far.
We have a docker working now which can also actively do “things”.
Limit is the amount of API commands available today, but we’re definitely using LLMs to not only support writing code but also analyse data in realtime, along with assisting answering complex question to less technical people.
This would be a great idea. One other competitor has announced this.