Arming from Apple Home now works the same as arming from the app. • Fixed: on some panels, arming Away from Apple Home could end up in Day mode. Apple Home and scripts now use "Programming" code to arm the panel. • Guides updated for 4.12. • Bug fixes and stability improvements throughout the app.

Helper restarts itself if it crashes. • Everything Helper does needs for Myro:Home Helper to be running. HomeKit, push notifications, the API, scripts, logging. If it quits at 2am you probably won't find out until something at home doesn't happen. • Now it comes back on its own a few seconds later. Crashes are rare, and the last few releases made them rarer, but the Mac running Helper is usually somewhere nobody looks. • You don't have to turn this on. It's part of "Start Myro:Home Helper automatically when logging into this Mac". There's an (i) beside that checkbox now if you want the details. • Quitting Helper with Cmd-Q keeps it closed, and so do logging out and shutting down. Also, a Force Quit looks like a crash, so Helper will start again if that occurs. • If it keeps crashing it gives up, after five tries in an hour, waiting longer each time. For example, restarting doesn't fix a full disk or some other system issue. • All four guides updated for 4.11. • Bug fixes and stability improvements throughout the app.

Clearer answers when something is wrong: bad keys say so, and one glitch is no longer a warning. • A rejected encryption key now tells you. When the controller refuses the secure session — the sign that the keys entered do not match the panel — the controller's row and the status line read "Invalid Keys", the log records it as a critical error, and the connection-status push says the keys were rejected instead of going out blank. Previously this failure was silent: the row simply read "Disconnected" and nothing explained why. • Ethernet Quality no longer treats a single glitch as a pattern. One unexpected disconnect and the reconnect that follows it were being counted as two incidents, enough to raise the concern level to Medium on their own. The level is now based on episodes rather than rows, recent episodes count for more than old ones, and it only rises to Medium when trouble is recurring or its frequency is increasing — one or two isolated events stay Low. • Ethernet Quality now uses evidence it already had. An episode that happened while your other local bridges were also unreachable is treated as a wider network event, not a fault on the controller's link, and the assessment says so in plain words rather than reporting nothing. • The Ethernet Quality chart now draws events as hourly bars instead of a line. A quiet week no longer reads as a flat line with one-pixel spikes, and a drop followed by a failed reconnect in the same hour shows as one stacked bar rather than two overlapping marks. • The HTTP API's Ethernet Quality response adds a one-sentence reason plus the episode, decay and frequency figures behind the level; every existing field is unchanged. • All four guides updated for 4.10, including how the new assessment works. • Bug fixes and stability improvements throughout the app.

Scripting that can hear the panel, remember things, and wait for fresh data. • Scripts can react to panel events directly — switch presses, macro buttons, Pro-Link messages, CentraLite switches, X10, Compose, camera triggers and UPB links all arrive through ScriptEvents.onOmniEvent, using the decoding Helper already had. Existing UPB link scripts keep working unchanged. • Persistent variables survive a restart — setPersistentVar, getPersistentVar and removePersistentVar keep small values scoped to one script and controller, so a counter or a last-run time is still there tomorrow. Files are written atomically and are not secret storage. • Ask for fresh object data, and know when it arrives — MyroController.refreshStatus requests a read-only refresh of known objects, getRefreshStatus reports progress, and ScriptEvents.onStatusRefresh fires when it finishes. Requests coalesce, a disconnect ends the ones in flight, and no commands are sent to your devices. • Helper also refreshes its object snapshots on its own after starting up and after reconnecting, so scripts, HomeKit and the API see current state sooner. • stopSound now actually stops a sound. Playback is tracked per script and controller, so a script can stop what it started, replace a sound of the same name, and never silence another script's. • Thermostat setpoints can move by a relative amount — adjustHeatSetpoint and adjustCoolSetpoint shift a setpoint without a script reading it first. • showCamera puts a camera on an HAI touchscreen. • Fixed: the relative cool adjustment carried the heat command's number, so asking to raise or lower cooling would have moved the heat setpoint instead. • All four guides are updated for 4.09, and the build now refuses to bundle a guide whose contents do not match its source. • Bug fixes and stability improvements throughout the app.

Your apps can now reach Helper from outside the house. • Reconnect over a VPN — Bonjour cannot travel beyond your local network, so an app away from home could never find Helper, even though Helper answers on every address the Mac has, including its Tailscale or other VPN address. Helper now tells an app every address it answers on, and the app remembers them. Open the app once on your home network to let it connect, and from then on it reconnects from anywhere without any setup. • Two new Getting Started sections — "Reaching Helper Away From Home" explains the one-time local connection and why the order matters, and "Remote Access with Tailscale" walks through setting up Tailscale on this Mac and on your phone. Nothing is exposed to the Internet and no router ports are opened; Helper needs no configuration, because it already answers on every address this Mac has. • Bug fixes and stability improvements throughout the app.

Your Omni, now something an AI assistant can actually operate — plus four rewritten guides. • Myro:Home Helper MCP server — connect Claude Desktop, Claude Code or any MCP-capable assistant to your controller through the Helper. Ask what state a zone is in, dim a light, set a flag, check the connection's health, or have it write an automation for you, all through clearly described tools rather than guesswork. Install with one config block: uvx myrohome-helper-mcp. • A new MCP Guide walks you through the whole setup — creating the right API key, installing uv, configuring Claude Desktop or Claude Code, confirming the connection, and what to ask once it works. • A new Getting Started guide covers connecting your controller, the first synchronization, and choosing the services that make sense for your home. • The Scripting and HTTP API references have been rewritten and are now generated from maintainable sources, so what you read matches what the app does. All four guides open from the Help menu, and the build refuses to produce them if a route, a scripting method, or a version number has drifted. • HTTP API can read as well as write — new routes list every object with its decoded state, fetch one object, return recent log lines, expose controller and audio-source metadata, and set flags and outputs directly with the correct command rather than through the unit route. • Sensitive API access is opt-in per key — direct arm/disarm and remote script management are separate capabilities chosen when a key is generated and fixed for its life. Security requests still need a valid panel user code, wrong codes lock the key out, and every attempt is logged with the address it came from. • Expanded JavaScript automation — scripts can now read complete controller and object snapshots, control thermostats, locks, blinds, outputs, UPB Links, user settings and area lights, arm and disarm without storing a panel code, inspect audio-source metadata, and react to flags, outputs, blinds, sensors, user settings and access-control events. • AI-assisted script authoring — an assistant can inspect your existing automations and real device names, validate new JavaScript without running it, and save or enable it only when you ask. A compact API reference, small enough to paste into any chat, is one click away in the Scripts window. • Your edits are safe from outside changes — if an assistant or another app updates a script you have open, the editor says so and keeps your draft, with a choice to reload theirs or keep yours. • Clearer Ethernet Quality guidance — ACK timeouts are explained in context, with a low, medium or high link-concern level based on recent recurrence rather than lifetime totals or a single transient event. • More accurate zone status in the API and scripts, with refreshed zone data after connecting to prevent outdated or incorrectly interpreted states. • Improved Omni-Link decoding for grouped unit updates and larger controller configurations. • Graphs page shows live controller status — a connection indicator in the header refreshes every ten seconds, any messages the controller is displaying appear beneath it and can be cleared from the page when opened from the Helper, and the whole page now lays out properly on phones and tablets. • Vestaboard messages that fit themselves — scripts, the HTTP API and the MCP server can now hand the Helper plain text and it wraps to the board's width and centers it top to bottom, so the same message looks right on a Flagship or a Note without picking rows. Send an exact grid of character codes for color tiles and art, or keep addressing rows one by one exactly as before. • Bug fixes and stability improvements throughout the app.

Scripting: Ethernet Quality, zone details, a dark editor, and more... • Ethernet Quality for scripts — MyroSystem.getEthernetQuality() and ScriptEvents.onEthernetQuality give your scripts the same facts as the graph: whether the controller link is up, error counts since launch, and whether this Mac and your other bridges still have the network. Publish it over MQTT, call an API, or send a notification when the controller is unreachable but the rest of the house is fine. MyroSystem.isControllerConnected() answers the simple question. • Zone status in scripts — a zone in an armed area used to report "bypassed" whether it was bypassed, restored, or merely open. Zone handlers now also receive a details object with the raw status byte and each field named, so a restore is simply bypassed going from true to false. • "*" for every handler — event handlers that take a zone, unit, audio zone, lock, thermostat or area name now accept "*" to match all of them; the callback still receives the specific name. • Script editor follows Dark Mode. • Copy for AI (compact) now includes every event handler's signature, so an AI assistant can write onZoneChange and the Lutron handlers without the full guide. • The Helper now announces itself on your local network, so Myro:Home on iOS can confirm it is running from the About screen. • HTTP API: set flags and outputs directly — the new flag route (/helper/{controllerID}/flag/{number}/{value}) takes 0–255 with the proper flag command, and the output route switches an output on or off. Setting a flag through the unit route capped at 100 and turned 100 into 1. • "Alert when bypassed" fired for every zone in an armed area, because armed was mistaken for bypassed. It now fires only for a bypass. • MyroSensors.isDataFresh() now means "confirmed during the current controller connection": a value stays fresh while the connection is live even if it has not changed, and reads false after a disconnect or restart until a new status arrives. Scripts that treated it as "changed in the last five minutes" should re-check their logic. • Bug fixes and stability improvements throughout the app.

Ethernet Quality: a record of your controller connection's health. • Ethernet Quality — the Helper now keeps an hourly record of the health of its connection to your controller: corrupted messages, dropped and half-open sessions, and failed connection attempts, together with whether the rest of your network was reachable at the time. An intermittent connection is hard to diagnose after the fact; a record kept from before trouble starts is what makes it possible to tell a panel's failing network hardware apart from cabling trouble or the network itself. • View the trend under Graphs as "Controller Ethernet Quality". The chart notes plainly whether recorded events happened while the rest of your network stayed healthy — the distinction that matters when deciding what to fix. • On by default and inexpensive — one small sample per hour. Turn it off per controller under Logging → Data Log Event Types. • Bug fixes and stability improvements throughout the app.

Objects that change on your controller, Sonos speakers that move, and steadier HomeKit. • Objects that change on the controller are handled properly. If your controller reports a different object under a number you had renamed or marked as a favorite — which happens when Lutron integration IDs are re-indexed — the Helper no longer carries the old name, favorite and HomeKit settings over to the new device. A unit whose control type changed also drops the display state that belonged to the old hardware. • Objects you delete on the controller now disappear from the Helper after a sync, instead of lingering as items that can never be controlled. If a script still refers to one by name, the Helper tells you which script and which object, so a rename on the controller doesn't quietly break an automation. • A Sonos speaker that changes IP address is found again. Announcements used to fail silently when a speaker's DHCP lease moved, with the speaker plainly working in the Sonos app. The Helper now recognizes the speaker by its room name, uses the address it is actually on, and remembers it. Group announcements on older S1 systems are addressed to the speaker that plays for the group. • Fewer interruptions while HomeKit is in use. Syncing or pruning objects while HomeKit was reading them could stop the Helper; it now passes over anything that has just been removed. • One script handler can now cover both Lutron systems. onLutronDeviceChange and onLutronButtonPress fire for LEAP bridges and RA2 repeaters alike, so a home with both no longer needs two copies of the same script. Existing protocol-specific handlers are unchanged. • Log and graph improvements. Exported log rows are terminated correctly so they import cleanly, opening a graph on a large sensor log uses far less memory, and log rotation no longer removes an old archive before the replacement has been written. • The Scripting Guide adds a section on RadioRA2 keypads, including why a keypad added to the Omni as a unit does nothing and how to make one work from the Helper. • The AI scripting reference now names your Pico and keypad remotes, and lists the button numbers each one sends. They were missing before, so a script written against the reference could listen for a remote by a name that did not exist and quietly never run. • Bug fixes and stability improvements throughout the app.

No more false phone-line alerts. • Fixed a false "Phone Line Dead" notification that could follow a phone call. When the controller reported two events at once — the line going back on hook and a program running, say — the second was misread as a dead phone line. Events are now read correctly, and camera-trigger events are no longer mistaken for phone-line events either. • With phone-line logging turned on, the Log records the exact event code the controller sent with each phone-line event, so a false dead-line report can be told from a real one. • A phone line reported dead while a call is in progress is no longer treated as a fault — the controller sends that same report when the line is simply in use. Dead-line alerts now follow the controller's own trouble status, so you are told when the line is really down and left alone when it isn't. • Corrected another form of the thermostat reading glitch. Myro:Home Helper already recognized when a HAI thermostat sends a temperature or setpoint about 115 degrees too high for a single update; the same controller quirk can also send one about 58 degrees too high, which is what a thermostat sitting near 75 degrees shows. Both are now recognized and corrected, so the current temperature, setpoints and outdoor temperature stay accurate — and so do your graphs and anything that reacts to them.

Fixes launching on Intel Macs. • Version 4.01 could not open on Intel-based Macs ("Check with the developer to make sure Myro:Home Helper works with this version of macOS"). The app's bundled frameworks were built for Apple silicon only; 4.02 ships universal builds of all of them. Apple silicon Macs were not affected. Everything new in 4.0: Version 4.0 — MQTT, Sonos announcements, and scripting that knows your devices. Your Omni controller can finally join the rest of your smart home — and speak up when something needs you. • Sonos announcements. Have your scripts speak through your Sonos speakers: "Garage door still open," "Water leak in the basement," "Someone's at the front door." Announcements duck the music and put it right back, so nothing you're listening to gets stopped. Click Discover and every Sonos room on your network turns up in a list to tick — address any room by name, or a whole group at once. • Trigger scripts from MQTT messages. Subscribe with topic wildcards, and JSON payloads arrive already parsed — no boilerplate. Pair it with Sonos and any sensor on your broker becomes a spoken alert. • Publish from scripts with the new MyroMQTT commands, including retained topics. Helper also announces its own online/offline availability, ready to drop into Home Assistant. • Set up one or more brokers under Add-ons ▸ MQTT: passwords kept in the Keychain, a one-click connection test, and automatic reconnection that restores every subscription. • Play your own sound files through Sonos too — a chime, an alarm, a recorded message — from a folder the app opens for you. • Wired into a whole-house or paging amplifier? Tick Chime and set a lead-in, and announcements start with a tone and a pause so the amplifier is awake before the first word. Works on older Sonos systems as well as current ones. • Three ready-made script templates to start from — two for MQTT, one for spoken Sonos alerts. Scripting got easier too. Type Unit. (or Zone., Thermostat., Flag., and more) in the script editor and pick from your controller's actual device names — no more hunting for exact spelling, and no more scripts that quietly do nothing because a name was off by a space. Custom names you've given devices work in scripts as well. The Execution Log now shows just the script you're looking at, updates live as its handlers fire, and copies to the clipboard as text — paste it into an AI along with the script and ask what went wrong. And a Lutron Pico no longer needs an assigned load to trigger a script: enroll it in the hub, name it, and press. Scripts can now also react to the controller's own text messages, and show or clear them. Sensor graphs are smoother and more honest. Auto-refresh now fetches only new readings instead of the whole history, the chart updates in place without flashing, and outlier removal looks at every series — setpoints and outdoor temperature included — marking each set-aside reading with a red cross you can hover, rather than silently hiding it. And a quirk in HAI thermostats that reports a temperature 115° too high for a single moment at a state change is now recognised and corrected as it arrives, so it never reaches your scripts, HomeKit, or your log files. And grab an AI-ready copy of the whole scripting reference from the Help menu — now including your real device names — then paste it into the AI assistant of your choice and just describe the automation you want.

Version 4.01 — no more freezes from scripting. • Fixed a freeze (spinning beachball) that could occur when a script sent to a Vestaboard that didn't respond, and a rarer freeze that never cleared when a script re-ran while the controller was reconnecting. Scripts and their event handlers now run entirely off the app's main thread, so nothing a script does can hold up the window. • A script or event handler that runs longer than 30 seconds is now stopped, and the message names the script. Previously a script stuck in a loop held up every other script for that controller until the app was restarted. • Saving a script now checks it without running it. Validation used to execute the whole script, so a save could send real commands to your controller. • Editing MQTT brokers no longer pauses the app while the old connections close.

Version 4.0 — MQTT, Sonos announcements, and scripting that knows your devices. Your Omni controller can finally join the rest of your smart home — and speak up when something needs you. • Sonos announcements. Have your scripts speak through your Sonos speakers: "Garage door still open," "Water leak in the basement," "Someone's at the front door." Announcements duck the music and put it right back, so nothing you're listening to gets stopped. Click Discover and every Sonos room on your network turns up in a list to tick — address any room by name, or a whole group at once. • Trigger scripts from MQTT messages. Subscribe with topic wildcards, and JSON payloads arrive already parsed — no boilerplate. Pair it with Sonos and any sensor on your broker becomes a spoken alert. • Publish from scripts with the new MyroMQTT commands, including retained topics. Helper also announces its own online/offline availability, ready to drop into Home Assistant. • Set up one or more brokers under Add-ons ▸ MQTT: passwords kept in the Keychain, a one-click connection test, and automatic reconnection that restores every subscription. • Play your own sound files through Sonos too — a chime, an alarm, a recorded message — from a folder the app opens for you. • Wired into a whole-house or paging amplifier? Tick Chime and set a lead-in, and announcements start with a tone and a pause so the amplifier is awake before the first word. Works on older Sonos systems as well as current ones. • Three ready-made script templates to start from — two for MQTT, one for spoken Sonos alerts. Scripting got easier too. Type Unit. (or Zone., Thermostat., Flag., and more) in the script editor and pick from your controller's actual device names — no more hunting for exact spelling, and no more scripts that quietly do nothing because a name was off by a space. Custom names you've given devices work in scripts as well. The Execution Log now shows just the script you're looking at, updates live as its handlers fire, and copies to the clipboard as text — paste it into an AI along with the script and ask what went wrong. And a Lutron Pico no longer needs an assigned load to trigger a script: enroll it in the hub, name it, and press. Scripts can now also react to the controller's own text messages, and show or clear them. Sensor graphs are smoother and more honest. Auto-refresh now fetches only new readings instead of the whole history, the chart updates in place without flashing, and outlier removal looks at every series — setpoints and outdoor temperature included — marking each set-aside reading with a red cross you can hover, rather than silently hiding it. And a quirk in HAI thermostats that reports a temperature 115° too high for a single moment at a state change is now recognised and corrected as it arrives, so it never reaches your scripts, HomeKit, or your log files. And grab an AI-ready copy of the whole scripting reference from the Help menu — now including your real device names — then paste it into the AI assistant of your choice and just describe the automation you want.

Connection and notification reliability. • Improved responsiveness when communicating with your controller — commands and status updates no longer wait behind other app activity. • Fixed a rare crash that could occur while reconnecting to the controller. • Behind-the-scenes updates to keep your security alerts and Live Activities delivering reliably.

Notifications tab improvements. • Added download buttons next to each app in the Notifications tab — including Myro:Home Pro — making it easy to get the right companion app on each of your devices.

Scripting engine stability hardening. • Fixed several rare crashes that could occur while scripting was enabled, especially with event handlers (e.g. ScriptEvents.onZoneChange) firing during frequent zone status updates. • Script event handlers and timers now execute on a dedicated serial queue for improved reliability. • Test push notifications now respect the per-controller Myro:Home / Myro:Home Pro checkboxes, matching real alerts.

Lutron level updates fix. • Fixed Lutron device levels not refreshing live. In Lutron → Devices, the Level column now tracks your lights in real time as they dim and brighten, instead of only updating when you disable and re-enable the bridge.

Audio zone control, UPB Links in HomeKit, and Lutron fixes. • Audio zone control. You can now control multi-room audio zones — power, mute, volume, source select, and transport (play, pause, stop, next, previous) — both from the HTTP API and from scripts using the new MyroAudio scripting object. A ready-made "Audio Zone Control" script template is included, and the API and scripting guides have been updated to document every command. • UPB Links in HomeKit. You can now expose UPB links to HomeKit and Home Assistant as on/off switches. In the Object Editor, right-click and choose "Add UPB Link…", give it a number and a name, then enable HAP. Note: the controller can't report UPB link status, so the switch state is best-effort — it follows link activity and starts off after the Helper restarts. • Lutron device levels now display correctly. The Lutron → Devices list showed every dimmer at 0% even when the lights were on. The Helper now reads the real level from your Lutron system when it connects and keeps it updated live as levels change. • Sortable Lutron device columns. Click any column header in Lutron → Devices — Name, Protocol, Type, Area, Reference, or Level — to sort the list, and the order is preserved as levels update. • Object Editor: double-click a device's name to jump straight into editing its Custom Name, instead of clicking into the Custom Name field separately.

Control Omni units from other systems. • Added an HTTP API command to set an Omni unit on, off, or to a dimmer level (0–100). This lets external systems — such as a Universal Devices ISY or Home Assistant — control your Omni lights and switches over the network, and keep other ecosystems in sync. The command uses your existing API key and follows the same style as the other Helper API endpoints.

• Fixed humidity readings for auxiliary humidity sensors sent to HomeKit. On controllers set to Celsius, the humidity percentage was converted as if it were a temperature, producing values that were far too low (for example, 50% could show as 10%). Humidity now reads correctly regardless of the controller's temperature format. • Enabling HomeKit on an auxiliary humidity sensor now automatically selects the Humidity type, so the right sensor shows up in HomeKit without also opening the type menu. An explicit Temperature/Humidity choice you've already made is never changed.

HomeKit stability and temperature/humidity support. • Fixed a crash that could make Myro:Home Helper quit immediately on launch when certain HomeKit-enabled sensors reported in. It most often showed up with temperature or humidity sensors that the OmniPro exposes as both a Zone and an Auxiliary Sensor. • Auxiliary temperature and humidity sensors can now be sent to HomeKit. In the Object Editor, enable HomeKit on an Auxiliary Sensor and choose Temperature or Humidity, and the real reading (not just a binary on/off) shows up in HomeKit.

Lutron scripting and startup fixes. • Fixed Lutron-triggered scripts not running. Script event handlers that respond to a Lutron change — a wall dimmer, Pico remote, keypad, or the Lutron app moving a light — now fire correctly. Previously these handlers could be set up without error but never actually ran, so one-way mirroring (for example, "when the Lutron light changes, match an Omni unit to it") silently did nothing. Button-press and UPB Link handlers are fixed by the same change. • Lutron bridges now connect automatically at app launch. Previously a bridge wouldn't connect until you opened Add-ons ▸ Lutron ▸ Bridges, which meant after a restart an event script could be listening with nothing connected. Restart-and-recover now works on its own, with no manual steps.

Scripting Improvements: • Script templates: the "Create Script" button is now always visible at the bottom of the template panel — previously it could be scrolled off-screen below the code preview. You can also now copy a template's generated code to the clipboard.

Lutron RA2 improvements. • Omni units configured for RA2 protocol now report correct 0–100% levels to scripts and the unified Lutron API, instead of the raw 100–200 encoded values. This fixes bidirectional Omni ↔ Lutron bridge scripts where dimmer levels were showing up as 101% or 150%. • You can now set a custom RA2 integration (LIP) password directly in the Lutron tab — click the lock icon in the Auth column. Previously only the default password was supported.

Scripting engine stability fix. • Fixed a rare crash that could occur in long-running scripts on macOS 26.5 (Tahoe) when an event handler's closure referenced one of the built-in objects (MyroSystem, MyroLutron, MyroAPI, etc.).