Panel Watch
See what the panel is reporting, without standing in front of it.
A trouble signal tells you a zone number at a moment in time. What it does not tell you is what that point is actually called, what every other point is doing, what else the panel has wrong with it, or whether any of it is still true by the time somebody is deciding whether to dispatch. To answer that, somebody goes to the keypad. Panel Watch reads the panel directly over Bosch's Mode 2 automation protocol and puts two things on a screen: every configured point with its current state and the panel's own name for it, and a health line that names every condition the panel currently has set, in the protocol's own words. It runs unattended — a stateful protocol gateway holding a cached state for every panel — and the dashboard is a client of that state, so the answer is already there when somebody asks for it.
Open the deck · Download the PDFPanel Watch watches. What it does not do is tell anybody. It runs unattended and holds a current state for every panel, and then it sits on that state until a person asks — there is no alerting, no notification and no escalation in it, nobody is paged, nothing is emailed and no ticket is raised. It does not receive alarms, and it replaces nothing your central station does. Watching is not the line; reaching outward is.
Who it is for
Estates running Bosch intrusion panels across more than one building, where the people responsible for them are not in the same building as the panel — a retail chain, a branch network, a campus with outbuildings. The people who feel it are whoever fields a trouble signal and has to decide what to do about it, and whoever is asked, once a quarter, which sites have a battery problem.
What it can see
Every configured point, with the panel's own name for it. Panel Watch asks the panel which points are configured, reads their text from the panel rather than from a spreadsheet, and lists each one with its state:
| State | What the panel is saying |
|---|---|
| Normal | At rest. The supervised resistor states are reported as normal, because they are not faults. |
| Open | The loop is open. |
| Trouble | Short or missing. The protocol distinguishes those two conditions and Panel Watch shows both as Trouble. |
The panel can also be asked directly for just the points that are not normal, rather than reading the whole list and filtering it.
A health line that names the condition rather than the category. Not "battery fault" — the panel reports these separately and Panel Watch lists every one that is currently set, worst first, or says that nothing is:
- AC power fail
- Battery missing, and battery low — different problems, reported differently
- SDI bus fail and point bus fail — a module on the bus, versus the point wiring
- Communication fail and phone line failure — the panel cannot reach where it reports to
- Failed to call RPS
- User code tamper
- Parameter CRC fail
- Log overflow and log threshold reached — housekeeping rather than faults, but shown, because losing events matters
The panel's own identity. Firmware version and revision, and the panel model, read from the same status response — so a fleet answer to "what is actually installed at that site, and on what firmware" does not depend on the asset register being right.
On demand, or on an interval. Refresh when you want it, or tick auto-poll and let it re-read on a set interval while you watch.
Export what is on screen. The current point list saves out as CSV, which is what you attach to the ticket or hand to whoever asked.
A panel view, and a dashboard over all of them. A panel view is one system: its points by the panel's own name for them, its areas, and every condition that panel currently has set. The dashboard is every panel — each one's current state on one screen, with faulted and unreachable panels called out, so an estate is something you look at rather than a sequence of site visits. It answers "which sites have a problem right now" in one look.
Up to a handful of panels this matters little. Somewhere past fifty or a hundred it stops being optional: opening four hundred panels one at a time does not answer which sites are faulted right now.
What it cannot see
A limitation is worth more than a feature list when someone is deciding whether to trust this, so:
- No history. The panel keeps an event log and Mode 2 exposes it. Panel Watch does not read it today, so it cannot tell you what happened last night, only what is true now. Reading it is the next thing we would add, and it is a build rather than a setting.
- No alerting. Covered above, and worth repeating because the word "monitor" invites the assumption.
- No API, and nothing pushed onward. Panel Watch has no integration surface today — its output is the screen and a CSV file. Fanning conditions out to a webhook, a ticket queue or a chat channel is a build, not a setting. Worth knowing which direction that build runs: Mode 2 is a socket-level binary protocol on a direct connection, so it is not something a third-party system calls into. Anything of this kind means us pushing outward.
- No alarm summary. Mode 2 has a dedicated alarm summary and detail, and neither is implemented, so Panel Watch cannot tell you an area is in alarm. What it does give you is every point that is not normal, which in practice is the list you want.
- No arm or disarm state on screen. This one is a display gap rather than a missing capability — the gateway reads area state from the panel already, disarmed through the several armed states. The panel view simply shows the point list and the health line, and does not put it on screen.
- No detail behind the bus bits. The protocol offers per-module SDI2 state, extended diagnostics, per-device and keyfob battery, and cellular signal strength and diagnostics. None of those are read today, so a bus fault tells you there is one, not which module.
How it goes
- Set the panel up to be asked. In RPS, set the automation device to Mode 2 and set an automation passcode. The panel needs firmware 3.03.014 or later.
- Put a Windows machine where it can reach the panels. All it needs is IP reachability to each panel on port 7700. Where your panel networks are already reachable from a central site, one machine covers the whole estate — there is no box per building. A machine at the site is only needed where its network cannot be reached centrally. Either way the panel itself is never internet-facing.
- Connect once and save it. Host, port, passcode and the certificate thumbprint to pin. On a licensed build these are embedded, so nobody types a passcode at a site.
- Read the panel from wherever you are. An outbound ConnectWise ScreenConnect session onto that machine, and the live view is in front of you.
What you get back
The question "is that still true, and what else is going on?" stops requiring a person at the keypad. An overnight signal can be checked before anyone is dispatched — the zone number becomes a named point with a current state, alongside everything else the panel has wrong with it. A tech who does go can be told what to expect rather than finding out on arrival. And a question that is normally a site visit each — which of these buildings has a battery that will not hold the panel up — becomes a look at all of them from the same chair.
Where it fits
| Panels | Bosch B-series and G-series over the Mode 2 automation protocol. The panel needs firmware 3.03.014 or later, with the automation device set to Mode 2 in RPS and an automation passcode set. Verified against a Bosch B9512G. There is no support for any other manufacturer. |
| Connection | TLS over TCP, default port 7700. Mode 2 is a socket-level binary protocol — a framed byte stream on a direct connection, not a web API with HTTP verbs and JSON — so anything built on it is written against the socket rather than called over the network by a third party. Certificate pinning is the intended setting; allowing an untrusted certificate is a logged LAN-only fallback and is off by default. The automation passcode is never written to a log. |
| Deployment | One Windows machine with the .NET 8 desktop runtime and IP reachability to each panel on port 7700. Where your panel networks are reachable from a central site, that one machine covers the estate; a machine at the site is only needed where its network cannot be reached centrally. The collector runs on one machine on a schedule; remote access to it is an outbound ConnectWise ScreenConnect session, so no site router needs an inbound rule and port 7700 is never forwarded. |
| Many panels | A panel view covers one system and its areas; the dashboard carries the current state of every panel on one screen, with faulted and unreachable panels called out. There is no history behind it — each collection pass replaces the last. Where the value is: at one panel or ten it matters little; past fifty or a hundred, opening them one at a time stops answering which sites are faulted right now. |
| Licensing | Each machine is licensed to its own hardware fingerprint, so every machine gets its own build, and the license carries one panel's connection details. Further panels are entered on the machine. Moving to a different machine means a new license. |
| Configuration | Bosch RPS-API is the complement for configuration and control. In Panel Watch that client is a scaffold: the transport and the error mapping are real, the endpoint routes are not built. Live status and health come over Mode 2 either way. |
| Support | Business hours as standard, with an optional premium tier. Coverage is set by the executed support addendum. |
Panel Watch is built, and was verified end to end against a live Bosch B9512G on 7 August 2026 and exercised during a customer trial. It is not in production at a customer today, and there is no reference call to offer. No uptime figure and no response-time promise appears anywhere in this material. Where a vendor product is named it is named because we integrate with it — not to imply a partnership, certification or endorsement.
See it on your own system
Book a thirty-minute walkthrough. Bring a panel you can reach — a bench unit is ideal — and we will connect to it and read it live: its points, their states, its health line, and its firmware. Pull the mains or the battery lead while we watch and see the health line name it. Have the firmware version and your RPS automation settings to hand, and we will tell you plainly what your estate would need. knightwatch.net
Knight Watch installs and services intrusion and access control estates as an integrator; the people who wrote the Mode 2 client are the people who have stood at the panel with a meter.