Removing the runtime while keeping ~/.openclaw left state that its own updates had migrated past the packaged release, and reinstalling seeded a release that refused to open it; even `openclaw update` refused. Removal now takes openclaw off PATH and leaves the runtime inside ~/.openclaw with the state, where a reinstall picks it back up; deleting ~/.openclaw still takes both.
Remove > AI refused whenever anything had Hermes's files open and told the user to close it and try again, and the thing open was Hermes: the agent in the terminal that choosing it as the default agent leaves running, the desktop app, a gateway. Those are the removal's to close. It now stops the gateway unit that upstream's `hermes gateway install` wrote, since that unit starts the runtime about to be deleted and would start it again the moment it was killed, then ends every process whose program lives in that runtime or in the package, and only then refuses over whatever is left, which is somebody else's: an editor on a skill, a shell sitting in ~/.hermes, a writer on the state database. Both are judged by the program, never by a later argument or by the home served, so an editor opened on a runtime file is the user's and a unit running a Hermes kept elsewhere is left alone whatever home it serves; for an interpreter the program is the script it runs, so a gateway started by hand as `python .../hermes gateway run` is found too. The refusal comes before anything is touched, so a run that stops leaves Hermes running as it was. A unit that will not stop still aborts the removal, as dropping the package would strand a live gateway on deleted code.
Co-Authored-By: Codex XHigh <noreply@openai.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Hermes is only ever installed through the app. Choosing it as the default agent used to stand aside for a hermes that worked but came from somewhere else, and to refuse one that predated seeded sessions; both left the machine on a Hermes that was not the app's, which is the one Omarchy prepares for in-app updates and hands the theme to. Now --check answers only for the app's own Hermes, and --now installs the app whatever answered to hermes before, saving the previous command aside as it always did for the app path. The desktop installer no longer adds the package itself, since the shared install does.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Choosing Hermes as the default agent built it through mise: a pipx environment with no checkout, so `hermes update` had nothing to move, and the only Hermes that could update itself was the one Hermes Desktop set up. Both paths now run the same setup. omarchy-install-hermes-cli installs the hermes-desktop package and runs upstream's installer from it, pinned to the packaged release and started on main, exactly as Install > AI did; omarchy-install-ai-hermes is that plus opening the app. The terminal, the default agent and the app share one runtime, and it updates itself.
--check answers whether --now has anything left to do, not merely whether a hermes runs: choosing Hermes from the menu asks first and opens a terminal only on a no, so a yes has to mean no minutes-long step would run where nobody can see it. With the app installed that means the runtime's own command, its completion marker and the seeded packaged app; a finished runtime whose command is gone, somebody else's, or its own but unable to run gets it back from upstream's path stage without bootstrapping again. Either way the command has to be the one PATH finds, because omarchy-agent runs bare `hermes` and Omarchy puts mise's shims ahead of ~/.local/bin; a command in the way is named rather than installed over. The modes are named outright because the app's launcher used to call this command with no arguments to reconcile a mise copy; a default of --now would turn every launch into an install. --check still refuses to run the retired wrapper, since running it built Hermes through mise, and a machine whose migration is pending can still have it on PATH.
Provisioning no longer writes the wrapper, Remove Preinstalls no longer looks for it, and the wrapper, the environment it built and what proves them Omarchy's are known to the installer alone: --retire-mise is the migration's whole job, and --now runs the same removal once the runtime installer has saved the wrapper aside, so a user who chose Hermes before their migration ran is not left with mise's shim answering `hermes`. Only the wrapper proves the environment is Omarchy's, at its path or in that saved copy, so the environment goes first and the wrapper last, judged by mise neither having it installed nor still requesting it; a removal that leaves either behind, or a listing that cannot be read, mise missing included, stops with the commands to finish by hand and leaves the migration pending. The migration that once installed the wrapper is kept as a no-op for late updaters, and one whose default agent was Hermes is told to choose it again.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Codex XHigh <noreply@openai.com>
The desktop background no longer plays videos. OWE owns video
backgrounds, and the shell layer stays empty behind one. The shell keeps
stills, which OWE hands back to it.
Remove the desktop video pause plumbing that only existed to stop an
unseen player: the lock, idle, and battery service lookups, the
per-output fullscreen check, the first-screen audio opt-in, and the audio
output in BackgroundVideo. The lock screen keeps its own silent playback.
Update the background tests, the manual, and the package note.
The background plugin now watches for the OWE daemon socket. While OWE is
running, the desktop yields video playback to it and the shell keeps
stills. The lock screen keeps its own playback.
This lets Omarchy cooperate with OWE without OWE editing shell.json, so
the engine can ship as a package.
Add a package-list note that owe-wallpaper-engine must be added once it is
packaged.
Three review findings on the update-hook boundary:
The pre-refresh-pacman hook had been moved after the refresh transaction
and, during a channel switch, deferred to the very end. That defeated the
hook's purpose: custom repositories and IgnorePkg entries were not in
place when the downgrade-capable -Syyuu ran. Run the hook where it used
to run, after the package config is re-synced and before the transaction,
but cold: revoke the timestamp, run it behind the no-update wrapper with
the caller's original PATH, and revoke again before continuing. Every
later privileged command authenticates with --no-update, so a detached
child left by the hook has no reusable timestamp to wait for. Channel
switching hands the caller's PATH to the refresh the same way the updater
receives it, and no longer defers or re-runs the hook.
Stay Awake was released before AUR builds, hooks and mise, so the machine
could sleep during the longest part of an update. Releasing the inhibitor
needs no privilege because the held command already dropped to the user,
so stop it after mise and before the reboot prompt, as before.
A packaged channel destination cannot be inspected before its package is
installed, and a transaction can replace the running tree with a release
that predates the command-scoped wrapper; from then on a bare sudo would
resolve to /usr/bin/sudo and publish a timestamp, and the destination's
own updater authenticates the same way. The switch used to abort only
after the packages had changed, with generic rerun advice. Now it checks
for the wrapper after each transaction before any further privileged
step, completes what it safely can, and stops cold with instructions to
run that release's update from a fresh session instead of launching it.
Boundary tests pin the hook between the config copies and the transaction
with a cold timestamp on both sides, the older-destination stop with its
guidance and no launched updater, the new inhibitor position, and the
post-update hook staying unreached on failures and signals. Docs, the
manual and the sample hook describe the restored timing.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Upstream now runs every Omarchy-owned pacman transaction through the
hidden omarchy-update-pacman helper so a mid-transaction systemd reexec
cannot kill it. Keep the deferred pre-refresh-pacman hook and the
command-scoped sudo wrapper, and call the helper from the refresh and
channel commands; the wrapper still applies to the helper's own sudo.
The sudo boundary fixture copies the helper into its root and runs a
systemd-run stand-in that execs the wrapped pacman step in place.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Remove unsafe project bin PATH injection
* Cover customized unsafe Mise paths
* Revoke legacy Mise Work trust
* Harden legacy Mise trust cleanup
* Preserve ignored Mise Work configs
* Scope Mise path cleanup to env
* Accept paranoid Mise ignore marker
Reported-by: infosec-us-team
The repo moved to the omacom org. GitHub redirects the old URLs, but
`omarchy channel set dev` was still cloning from basecamp/omarchy, which
left every dev checkout with a stale origin remote that confuses gh
(pr create fails with "No commits between omacom:quattro and
basecamp:<branch>"). Update the clone URL, the quattro upgrade tarball,
the update-confirm release link, the systemd Documentation link, and
the manual.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LSFKDatumRZHB8zqk5CP5C
Follows the ChatGPT flow: the Install > AI entry runs
omarchy-install-ai-claude in a floating terminal, which installs the
claude-desktop package (Anthropic's Linux desktop beta, repacked from
their Debian repo in omarchy-pkgs) and opens the app. Remove > AI
drops the package along with ~/.config/Claude and ~/.cache/Claude,
the Electron directories the desktop app owns, while keeping
~/.claude, ~/.claude.json, and ~/.cache/claude-cli-nodejs: those
belong to the Claude Code CLI, which ships in its own package and
survives this removal, just as the ChatGPT remover keeps the Codex
CLI.
The menu mark is a new U+E90E glyph in the Omarchy icon font, from
Simple Icons' Claude mark, so it reaches desktops through the next
omarchy-settings release.
Co-Authored-By: Fable 5 <noreply@anthropic.com>
Co-Authored-By: Codex XHigh <noreply@openai.com>
Hiding the top bar and removing the window gaps are the two things you
do to give the screen entirely to your windows, and doing both took two
hands and two hotkeys. `omarchy toggle fullscreen desktop` does them
together.
It only leaves full screen when both halves are in it, so hitting the
hotkey with just the bar hidden (or just the gaps gone) pulls the other
half into line instead of flipping the one you already set.
Claude-Session: https://claude.ai/code/session_01JB9phxP56gnP7qSidkkUJE
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Add Muse Code as a default coding agent
Meta ships Muse Code only as a binary, so it installs from the AUR
(muse-code-bin) instead of mise, and a fresh install runs the muse login
browser flow in the install terminal before the agent opens.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0148qKzr366p2Ubu2igCLvPg
* Refine Muse Code menu and prompt forwarding
* Install Muse Code from OPR
* Install Muse Code through mise's HTTP backend
* Preinstall the Muse mise stub
* Use the shared Muse installation flow
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: David Heinemeier Hansson <david@hey.com>
* Add Cursor CLI as a coding agent choice
* Launch Cursor CLI through its agent subcommand with --trust
Cursor CLI still dispatches a one-word prompt that names one of its
subcommands (update, login, help) even after a bare --, so name the agent
subcommand outright and pass the prompt behind -- there, where it also
keeps a prompt starting with a dash from being read as an option. --yolo
only auto-allows commands; the workspace trust dialog is skipped only by
--trust, and a launcher that must not stop to ask needs both.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Leave an official Cursor CLI install alone
Cursor's own installer symlinks ~/.local/bin/cursor-agent, the same path the
mise wrapper takes. The migration now installs the wrapper only when no
cursor-agent command exists, and Remove Preinstalls deletes the path only
when it holds the wrapper omarchy-mise-install wrote, the way the Hermes
wrapper is handled. Selecting the agent treats an executable at that path
other than the wrapper as the user's own install and skips mise, since the
mise shims precede ~/.local/bin on PATH and a mise copy would only shadow
it. The wrapper resolves through mise's registry, which lists cursor-agent
from 2026.8.15 on.
The tests write a real cursor-agent stub before Remove Preinstalls runs,
cover the preinstall opt-out for the new migration, and check that a
symlinked official install survives removal and selection alike while a
dead file at the same path still installs.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Give Cursor its brand mark and one name in the menu
Add Cursor's mark to the Omarchy icon font as U+E90D and point the agent
entry and both editor entries at it, so one brand is drawn one way across
the menu. Label the agent entry "Cursor CLI", the name the command and the
manual already use, and spell it the same in the migration and the tests.
Append the manual row after the others at the standard width.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Keep an official Cursor CLI install through a user re-provision
User setup writes every mise wrapper unconditionally, which is fine on a
fresh install but replaces the symlink Cursor's own installer leaves at the
same path when omarchy-provision-user runs again with --force. Guard that
one line the way the migration does.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: John Cavanaugh <59479+cavanaug@users.noreply.github.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* Add native video wallpaper support
* Pause video wallpapers while a fullscreen app is focused
* Sample one frame when a video background sets the bar text colour
A video wallpaper made the transparent bar's colour sampling decode the entire file. ImageMagick's video delegate runs ffmpeg with no frame limit, so a twenty-second 1080p background took 11.3s of CPU where one frame takes 0.14s, and it did that on every theme change.
The result was unusable anyway: a multi-frame input emits one value per frame, which the single-value match then rejected, so transparent bars silently fell back to the plain text colour on every video wallpaper. Selecting frame zero fixes the cost and the colour together, and fixes animated GIFs, which had the same bug.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Load wallpaper video lazily, and without an audio output
Three costs the still-image path should never have paid.
BackgroundMedia imported QtMultimedia at file scope and was instantiated on every output, so the module and its audio dependency closure mapped into every shell process whether or not a video was ever shown — measured at +2.72 MiB RSS. Moving the element into its own file behind a Loader that takes a URL defers the whole import: an inactive loader maps none of it, an active one maps all 25 libraries. An inline Component cannot defer that, because the type has to resolve when the file compiles.
Qt's Video convenience type always builds an AudioOutput, and `muted` only aliases that sink's volume, so every monitor decoded an audio stream it would never play and opened an audio client for it. A bare MediaPlayer with no audio output spawns no QFFmpeg::AudioR, QAudioContext or PWDevMon thread, and plays files with no audio track just the same.
The shared image also turned mipmapping on, which the desktop background never had. A full mip chain is about a third more texture memory — 10.6 MiB extra at 4K, per output — for a wallpaper drawn at its own size.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Stop wallpaper playback while the session is locked or screensaved
Playback stopped only for a focused fullscreen window. Locking the session did not stop it, and the lock screen starts a player of its own, so an N-monitor desktop reached 2N decode pipelines the moment it locked — and stayed there, because a display blanked for idle stops being presented but does not stop Qt's FFmpeg engine, which drives its own clock. A laptop locked with the lid shut decoded video until the battery ran out.
The lock and idle services already know both states, so the background service takes the shell reference the loader offers it and reads them. Looking a service up by id needs the registry to be reactive, or a background that loads before the lock service would bind to null and stay there.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Fan out video thumbnails narrower than single-threaded image jobs
The generator fans out one job per core, which was bounded because VIPS_CONCURRENCY=1 made each of them single-threaded. ffmpegthumbnailer leaves FFmpeg's automatic decoder threading on, so a folder of uncached videos put a codec thread pool on every core at once. Queueing video work separately keeps the still-image path at full width and gives the video path a quarter of it.
* Recognize a named video file as a theme preview
The backgrounds fallback beside it already picks videos, so a theme shipping preview.mp4 was the one case that still went unseen.
* Document video backgrounds in the manual
The manual described backgrounds as images only. Worth saying plainly that a video wallpaper costs far more power than a still one and that each monitor decodes its own copy, since neither is visible from the picker.
* Stop the lock screen's own playback once the displays go dark
Pausing the desktop wallpaper on lock only moved the cost. The lock screen builds a player per monitor of its own, so locking an N-monitor session went from N decoders to N rather than to none — and the lock service blanks the displays five seconds later without touching them, which is where a lock spends nearly all of its time. A laptop locked and shut still decoded video into a dark panel.
The service already owns both transitions, so it records whether the displays are dark and the lock view stops playback while they are. The manual said playback stops while the screen is locked, which was the same overstatement; it now says once a locked screen has gone dark.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Keep videos out of the lazy thumbnail path
A lazy row stands in with the media file itself until its thumbnail exists, and the picker draws that with an Image — which shows a picture and shows nothing for a video, with no reload once the real thumbnail lands. So the first open after discovering an uncached video showed a blank tile.
The same branch also spawns one generator per file immediately, before either queue is reached, and the theme switcher always asks for lazy thumbnails. That put the narrower video fan out on the one path that never used it: forty uncached previews meant forty ffmpegthumbnailer processes. Sending videos to the queue instead fixes the blank tile and puts them back under the cap.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Rebuild the theme preview cache after teaching it about video
Preview discovery changed what it recognizes, but its cache keys on theme directory mtimes alone. A theme that already shipped a video preview would keep whatever the old rules cached until something happened to touch the directory. Bumping the version rebuilds it once.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Drop an activeAudioTrack setting that never took effect
Qt's FFmpeg backend ignores setActiveTrack while no source is open, and the literal binding is not reapplied once the media loads and the tracks become known, so the line did nothing. What actually keeps the audio decoder and its client from ever being built is the absent audio output, which a file carrying an audio track confirms on its own: no QFFmpeg::AudioR, QAudioContext or PWDevMon thread appears without it.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Give up the blank state when a display comes back
The lock screen stops its wallpaper while the displays are dark, but it was tracking the blanking it asked for rather than the panels themselves. Opening a docked lid turns the internal panel back on without going through runWake, and so does a resume, which left a visible lock wallpaper frozen on one frame until the next keypress. A frozen wallpaper someone is looking at is worse than the decoding it saves, so a screen change gives the state up.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Time bound the video thumbnail generator
Routing videos through the queue means they are generated before the picker opens rather than behind it, which turned an unreadable or stalled file into a picker that never opens. ffmpegthumbnailer had no bound of its own and the drain waits for every job. A generator that gives up is already handled: the run reports failure, the partial file is removed, and the row drops out of the list.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Pause only the output a fullscreen window covers
The fullscreen test was global, so a game on one monitor stopped the wallpaper on every other one — including the ones still in plain view. That is the failure the lock work was careful to avoid, and it made the manual's claim that playback stops when nothing can see it untrue for the commonest multi-monitor case. A lock or a screensaver does cover every output, so those stay a single decision; fullscreen is now matched against the focused monitor, the way the bar already routes by output.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Kill a video thumbnail generator that ignores the timeout
Plain timeout sends TERM and then waits for a process that may never take it, which leaves the bound it was added for unenforced on exactly the stuck files it was meant to catch.
Co-Authored-By: Codex XHigh <noreply@anthropic.com>
* Pause video wallpapers in battery power-saver
* Fix paused video wallpaper source priming
* Skip snapshots for video background transitions
(cherry picked from commit 6f759538bfa76c2da03634e98ebfc2ebf63ec68e)
* Generate thumbnails for direct-scan videos
(cherry picked from commit 10fcca018a865dca311fb6863e8c8b0057291223)
* Remember a video the thumbnail converter rejected
A permanently unreadable video cost ten seconds of generator time on every
picker open before its row dropped, because nothing recorded the failure.
Both the menu image generator and the direct picker scan now leave a marker
beside the missing thumbnail, keyed like the thumbnail on the file's size
and mtime, so a repaired file starts clean. A timeout is left to retry, as
it may only have been a busy machine.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Follow the panels' real DPMS state under a locked video wallpaper
The lock screen stopped video playback when it asked for the displays to
blank, and resumed on input, but never checked what the panels did. A blank
that failed left a lit panel on one frozen frame, and a resume that turned
the same outputs back on played nothing until the next keypress.
Quickshell exposes no DPMS signal, so while a video is the locked wallpaper
the lock polls hyprctl and decides per surface from the answer. A wake or
blank request drops the last answer so its optimistic state applies until
the next poll confirms it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Pause a video wallpaper for the fullscreen window that covers it
The fullscreen check read the globally active window and the focused
monitor, so it only knew about the window that had focus. A fullscreen
window left on one monitor while focus moved to another resumed the
wallpaper decoding behind it, and with fullscreen windows on two outputs
only the focused one paused.
Each output's visible workspace reports whether a fullscreen window covers
it, and Quickshell flips that on the compositor's fullscreen event, so each
panel now decides from its own monitor's active workspace instead.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Reopen a video wallpaper a theme switch replaced behind its path
Two themes that both ship backgrounds/wallpaper.mp4 leave the current
background at the same path after a switch, so the displayed path never
changed and the running player kept decoding the old file from its open
descriptor. Stills go through the snapshot transition and survive this;
a video switch is instant and did not.
A forced switch onto the path already on show now bumps a reload counter,
and BackgroundMedia rebuilds the video player for it. A cache-busting query
is not an option there, since FFmpeg reads it as part of the filename.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Keep picker rows uncached while a rejected video is left out
Skipping a video with a failure marker let the picker cache its rows
without it, and cached rows are trusted on the directory's mtime alone.
A file repaired in place never touches that, so the marker's fresh key
was never consulted and the video stayed missing.
The generator now hands the marker back to the row loop, which drops the
row and leaves the rows uncached, so each open re-stats the file and a
repaired one is converted again.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Hand each background loader only its own kind of file
BackgroundMedia fed one URL to both the still loader and the video player.
On a switch from image to video the Image was handed the video's URL in
the moment before its loader unloaded, so Qt tried to decode the mp4 as a
picture and logged an unsupported format on every such switch; the reverse
handed the player a still to demux.
The still URL is now empty whenever the path is a video and the video URL
empty whenever it is a still, so a switch changes only the loader that
stays.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Stop a video wallpaper before tearing its player down
Switching from a video to a still destroys the BackgroundVideo item while
its player is mid-read, which FFmpeg reports as a failed open in the shell
journal on every such switch. Stopping the player on destruction lets the
demuxer wind down first, and the switch is quiet.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Play a video wallpaper's sound track from the first monitor
Video wallpapers were always silent: the player was built without an
audio output, since a muted output still decodes the track and opens an
audio client on every monitor. A video with music should be able to play
it.
The player now builds its AudioOutput only once the media reports a sound
track, so a silent file still opens no audio client, and only the first
screen's panel opts in, so a multi-monitor desktop does not layer copies of
the track. The output is muted while a paused player primes its first
frame, and the lock screen stays silent.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Keep a departing video player off the still's file
The switch away from a video still logged a cancelled open, and stopping
the player on destruction only hid it: stopping reports the media as
loaded, which the loaded handler answered by playing again. The real cause
was one evaluation pass. Both URLs derived from the `video` flag, which is
itself bound to the path, and QML updates the two in no fixed order, so
the video URL could evaluate against the stale flag and hand the player
the still for a moment. Its destructor then cancelled that open.
Each URL now tests the path directly, the source binding only applies
while the path is a video and restores nothing when it stops, and the
destruction stop goes away with the hazard it introduced.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Pin the audio wiring in the test and name the output in the manual
The audio assertion passed with the BackgroundMedia forwarding binding
removed, which would have left every wallpaper silent, and did not pin the
silent default or the first-screen selection. It covers all three now.
The manual said the sound track plays "from your first monitor", which
reads as routing to that monitor's audio device. It is the first monitor's
wallpaper that plays, through the default output.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Omabot <omabot@omarchy.org>
Co-authored-by: Codex XHigh <noreply@anthropic.com>
Co-authored-by: z8 <yam@kernelius.com>
Co-authored-by: David Heinemeier Hansson <david@hey.com>
Hermes Desktop installed under Install > AI kept its own palette while every other agent app retinted with the theme. Hermes' skin is its one theme unit for the desktop app, the TUI and the CLI, and its gateway watches the active skin file and broadcasts changes to every surface, so Omarchy publishes a skin named omarchy from a template on every theme switch and nothing Omarchy-specific goes upstream.
Activation goes through hermes config set, which writes the active profile's config and touches the skin so a running gateway repaints at once, and it only replaces Hermes' default skin so a choice made in Hermes stays. A theme switch runs that activation too when the desktop package is present and Hermes is still on its default, so a hand-over the installer missed is finished by the next switch; once the config names the skin a switch never starts Hermes. The desktop adopts a skin from a change broadcast rather than from the config it finds at connect time, and its first launch builds the runtime over minutes, so the installer starts --wait as a transient user unit that outlives the install terminal, activates once the runtime marker appears, republishes after the gateway is up, and reports to the journal. A migration hands the skin to existing Hermes Desktop installs through --activate, which also renders the skin for a theme applied before the template existed.
The generated file is validated before it is published, because Hermes parses it as YAML: only the name, a plain description and #rrggbb colours pass, so an unresolved palette key or a cloned theme's own hermes.yaml leaves the previous skin in place.
🤖 Generated by Fable 5.1 in Claude Code. Reviewed by Fable 5.1 code-review at high.
Follows the T3 Code / Grok Bot flow: the Install > AI entry runs
omarchy-install-and-launch, so picking it installs the perplexity
package on demand and launches the app when the install finishes.
Remove > AI drops the package along with the app's own config, flags
file, and rpc-server runtime cache, keeping the perplexity-* caches
that belong to Perplexity's other products. Like the Hermes remover,
it sets -u so an unset HOME is a refusal rather than rm -rf paths
rooted at /.
The menu mark is a new U+E90B glyph in the Omarchy icon font, so it
reaches desktops through the next omarchy-settings release.
Co-Authored-By: Fable 5 <noreply@anthropic.com>
Co-Authored-By: Codex XHigh <noreply@openai.com>
OpenClaw's desktop experience on Linux is its Control UI, served by the
gateway the openclaw package runs, so the Install > AI entry installs
the package and a web app launcher that routes through the new
omarchy-launch-openclaw: first launch hands off to OpenClaw's own
onboarding wizard, later launches start the gateway when needed and open
the dashboard's single-use browser handoff URL as an app window.
Remove > AI tears the gateway service down through OpenClaw's own
gateway uninstall (falling back to systemctl by hand), aborts rather
than dropping the package under a gateway that will not stop, and keeps
the user's agent in ~/.openclaw.
OpenClaw also joins Setup > Defaults > Agent through the same
agent_installer seam Hermes carries: its CLI is the pacman package
rather than a mise tool, so omarchy-install-openclaw-cli answers
--check/--now with pacman, and omarchy-agent runs `openclaw chat`,
seeding prompts through --message.
The menu mark is a new U+E90C glyph traced from the package's lobster
favicon; E90B stays free for the Perplexity mark still in flight on its
own branch.
The launcher recovers the gateway through `openclaw gateway install --force`
(unit not enabled: missing, or an install that died after writing it) or
`openclaw gateway start` (enabled but stopped), never `openclaw dashboard
--yes`: as of OpenClaw 2026.9.1 that defers to "the owning supervisor" in
both cases, and once the gateway is up it copies a one-time browser pairing
URL into the clipboard. The dashboard probe is bounded so an app-grid launch
cannot hang without a terminal to interrupt it. Removal treats only
systemd's own "inactive"/"failed" as a stopped gateway, so an unreachable
user manager aborts instead of dropping the package under a live process.
All of it verified against a real 2026.9.1 install.
Removal also takes down the node-host unit if OpenClaw ever installed one, and
asks (default no, only on a terminal) whether ~/.openclaw should go too, with
its size: the chats and credentials live there next to hundreds of megabytes
of plugin runtimes and cache OpenClaw downloads for itself.
Onboarding goes through omarchy-openclaw-onboard rather than bare `openclaw
onboard`: as of 2026.9.1 the bare command is the guided flow, which ends by
running a foreground gateway and handing off to a browser tab without
returning, so the install script never reached the app launch and no service
was installed. The helper runs the classic wizard (--flow quickstart
--install-daemon --skip-ui) as a background job that keeps the terminal as its
stdin, so its prompts render and take input as upstream draws them, and stops
it once the gateway answers: upstream leaves the wizard running after its
outro (only the TUI branch exits, and the model sign-in holds a socket open).
Every quickstart prompt precedes the service install, so that point is safe.
A gateway that never comes up after this run applies setup ends the wait as a
failure instead
of hanging, an already-running OpenClaw is left alone rather than mistaken for
this run's success, a gateway answering on the port is only this run's once its
process is the unit's own MainPID (an orphan from an
earlier run) is not mistaken for the service this run installs, and a signal at
the helper takes the wizard down with it.
Spencer Bull