Add the Perplexity desktop app to Install > AI
(cherry picked from commit f1b065c292)
(cherry picked from commit 680930915eaba8c3741980c56ad160a40c50bcb9)
Ask, default no, before Remove Hermes deletes the user's data
(cherry picked from commit 959e49dc52)
(cherry picked from commit 4e77199945a15ac194511b1af6317761fb813ba0)
Add OpenClaw as a desktop app and a coding agent
(cherry picked from commit eb56446c42)
(cherry picked from commit bde2584b5596412142d0a2039d44dd6a25949517)
Fix the Hermes CLI readiness probe, and tear the CLI down on uninstall
(cherry picked from commit f99d33a8dd)
(cherry picked from commit de739ea736563763150725ce33151001c5966add)
Add Hermes as a desktop app and a coding agent
(cherry picked from commit b71dcad96e)
(cherry picked from commit b27908369ed5fa95ca24dca2c68b895032d879d2)
Correct the local conflict parser consuming context beyond the selected hunk. Preserve all baseline entries and add only the selected T3 and removal scaffold rows.
(cherry picked from commit 473713413a9132375b6bf96627e8220f118ddf00)
Port only Remove > AI and T3 removal/tests from #7504 (023021ad2d). Keep Dictation in its baseline location and omit unrelated removers.
(cherry picked from commit 73480ffabc091b812a0cfbe7bc845799c6078a88)
t3code-bin is in the Omarchy repo now, so the menu can offer it the way it offers Cursor and Grok Bot: install the package, then launch the desktop entry it ships.
The mark is a trace rather than a download. T3 publishes no monochrome SVG — the app icon is a black rounded tile with the letters knocked out of it, and a tile flattens to a solid square once the menu recolors every path with the theme foreground. Tracing the lettermark out of that icon keeps the silhouette that actually reads.
The font is package-owned, so the glyph reaches a desktop through an omarchy-settings release rather than omarchy update. Until that release lands, a pulled checkout draws the entry with no icon.
🤖 Generated by Opus 5 in Claude Code.
(cherry picked from commit 260a729104)
(cherry picked from commit a863dc8555a51e62535f0edf28d4158d9d2e9fb4)
Exercise production JavaScript from #9485 and #9618 together with a synchronous Qt test double. Cover keepLoaded identity, detached authentication retention, refreshed manifests, cleanup eligibility, sticky classification, and clone lookups without the unrelated video-wallpaper services alias. Native QObject, PAM, and Wayland validation remains separate.
* Honor keepLoaded for services during plugin hot-reload
Plugin reload destroyed every service, including omarchy.lock, which drops the ext-session-lock client while Hyprland still holds the lock and surfaces the crashed-lockscreen fallback.
* Prove keepLoaded service survival with a fixture service
A fresh lock service also reports an empty lastEventAt, so comparing it
across the rescan passed whether or not the instance survived. A fixture
keepLoaded service whose in-memory marker is set before the rescan and
read back after can only pass when the same instance is still mounted.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Drop kept services whose plugin no longer declares a service
The _syncServices cleanup only asked whether the plugin was still
installed and enabled, so a kept service whose plugin dropped its
service kind or entry point kept running as a zombie until shell
restart. Apply the same eligibility checks used at creation, and hand
kept instances the refreshed manifest after a rescan.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* Cover omarchy.media in keepLoaded expectations; note kept services reload on restart
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---------
Co-authored-by: David Heinemeier Hansson <david@hey.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
(cherry picked from commit d3d23fddde)
* Quote install-app and install-font names like install-and-launch
* Quote the package list too, not just the display name
The display name was quoted but omarchy-pkg-add's own arguments were still interpolated into the bash -c string raw, so `omarchy install app Vim 'vim; id'` ran id. The list has to reach the helper as several words, so it cannot be quoted whole: it is split the way the unquoted expansion split it and each word is quoted on its own. Reading with -d '' keeps a newline-separated list intact instead of dropping every package after the first, which plain read -a would. install-font's package is singular and is quoted whole, and install-and-launch carried the same flaw.
Reported by acrogenesis in review of #7843.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Codex XHigh <noreply@openai.com>
* Test that install-font skips font-set when pkg-add fails
The hostile-package case was asserting the family still got set, which only held because the mock always exits 0. pacman would reject that name and the && chain would skip font-set.
* Keep the installers working when errexit is inherited
read -d '' always ends at EOF rather than on its delimiter, so it reports failure on every input. Under an inherited errexit the installers exited there and built no command at all.
Reported by Codex XHigh in review of #7843.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Codex XHigh <noreply@openai.com>
---------
Co-authored-by: David Heinemeier Hansson <david@hey.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Codex XHigh <noreply@openai.com>
(cherry picked from commit 625c4a1603)
Cherry-picked from quattro (7d58bb9a).
Stop world-writable Chromium and Firefox policy directories: create them
root-owned at 0755, purge non-root entries, refuse planted symlinks, and
write the browser theme colour through a passwordless helper instead of
a world-writable policy file.
Conflict resolution for v4-0-2:
- bin/omarchy-install-browser: dropped the `chromium)` case, which does
not exist on this branch.
- test/shell.d/default-apps-test.sh: dropped; the file does not exist on
this branch.
* Fix Codex usage collector approval policy
* Capture codex argv with boundaries in the scanner test
The stub joined its arguments with "$*", so the assertion compared one
flattened string and could not tell five arguments from fewer containing
spaces. Passing "-s read-only" and "-a on-request" as single arguments --
which codex rejects as an unexpected argument -- passed the test. NUL
separation and an array comparison keep the boundaries the assertion is
about.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Codex XHigh <noreply@openai.com>
---------
Co-authored-by: Omabot <omabot@omarchy.org>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Codex XHigh <noreply@openai.com>
(cherry picked from commit 4cd8a081cb)
The wildcard granted passwordless root for timedatectl set-timezone plus any trailing arguments, so -H/--host and -M/--machine reached the SSH and machine transports as root. Systemd 261 guards argv injection into ssh, but -H still drives root's SSH client at an attacker-chosen host, and the transport resolves its helper through PATH; only Defaults secure_path stands between that and a planted ssh running as root. Match the argument with an anchored POSIX ERE that admits exactly one timezone token (no whitespace, no leading-dash segment, no traversal component), so no second argument and no option can ever match. The sole caller, omarchy-menu-timezone, passes one list-timezones value and is unaffected.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Codex XHigh <noreply@openai.com>
(cherry picked from commit 0ae1694830)
Backport of the FIDO2 authfile fix (PR #7904 by @mdisec, merged to quattro as
23dab9ec) onto the v4-0-1 release branch.
pamu2fcfg wrote to /tmp/fido2 and the registration was then moved into place
with sudo mv. Any other local user can pre-create /tmp/fido2, and rename(2) does
not dereference the final component, so the privileged move installed the
attacker's symlink itself as pam_u2f's global authfile -- a file consulted by
sufficient lines in /etc/pam.d/sudo and /etc/pam.d/polkit-1.
The same move also carried the staged file's ownership into /etc, so on every
install to date /etc/fido2/fido2 is owned by the invoking user at mode 0644.
That needs no attacker, no race and no second account: anything running as that
uid can append its own credential and satisfy the machine's sudo prompt without
knowing the password.
The setup now creates a unique staging file as root beside the final authfile
and pipes pamu2fcfg into it, so root never reopens a caller-owned pathname,
rejects failed or empty enrollment output, publishes with an atomic mv -Tf,
cleans the exact staging file after every failure, refuses non-regular authfile
states, and installs root:root 0644. A migration repairs machines set up by
earlier versions by replacing the inode rather than chowning in place: a process
that already holds a writable descriptor on the legacy user-owned mapping keeps
it, so the repair has to leave that inode behind where PAM no longer reads it.
Clean cherry-pick: all six files are byte-identical to quattro, so merging
v4-0-1 into quattro resolves without a conflict. The migration's timestamp
(1787494718) is older than others already on this branch, which is harmless:
omarchy-migrate marks each migration by filename rather than tracking a
watermark, so this one runs on every machine that has not run this exact file.
test/shell passes: 195 files, including the three this adds -- the setup's
staging and failure paths, the migration's repair and unrepairable states, and
the removal's symlink handling. test/cli passes.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the input-device name fix (PR #8129 by @acrogenesis, merged to
quattro as 9285b19d) onto the v4-0-1 release branch.
Hyprland input-device and monitor names come from USB descriptors and hyprctl
output, so they are attacker-influenceable, yet the toggle and monitor commands
interpolated them straight into hyprctl eval and into generated Lua that
Hyprland re-executes on every reload. XF86TouchpadToggle is bound with
locked = true, so a malicious USB name reached Lua execution from the lock
screen as the logged-in user, and a persisted disable made it run on every
start. Publicly reported by Jorrit Jongma / Chainfire.
The disable is no longer executable Lua anywhere. The device name is stored as
plain-text data in a *-disabled-name sidecar and read back by a packaged module,
default/hypr/disabled-input-device.lua, on every reload; the live hyprctl eval
Lua-quotes the name and rejects control characters outright. The reload loader
excludes the two legacy filenames, so a leftover generated *-disabled.lua on a
not-yet-migrated install can never be sourced as code again, and a migration
recovers the device name from it and deletes it, sanitizing installs that ran
the vulnerable version. All four monitor scripts validate an output name against
a plain-connector-name pattern before writing it as Lua, closing the same latent
pattern in the siblings, and paths.lua treats a set-but-empty XDG_STATE_HOME as
unset to match the bash side.
Clean cherry-pick: all fourteen files are byte-identical to quattro, so merging
v4-0-1 into quattro resolves without a conflict. This branch ships no leftover
*-disabled.lua template of its own -- the tracked "disabled" files are the same
two quattro has -- so the migration is the only path that has to sanitize
anything here.
test/shell passes: 192 files, including the three this adds. The toggle suite's
public-PoC case passes here, as do the monitor scripts' accept/reject cases and
the XDG path cases. test/cli passes.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the DNS PATH pin (PR #8172 by @mdisec, merged to quattro as
4637735a) onto the v4-0-1 release branch.
omarchy dev link prepends a user-writable checkout's bin/ to sudo's secure_path
so privileged Omarchy commands resolve to the development versions, and that
reaches the subprocesses they launch too. This branch carries the same
passwordless grant -- etc/sudoers.d/omarchy-dns lets wheel run
/usr/bin/omarchy-dns Cloudflare, Google and DHCP without a password -- so the
packaged script ran as root while resolving bare helpers (dirname, install,
tee, rm, nmcli, systemctl, awk) through the caller's secure_path. Write access
to a dev checkout became arbitrary root execution, with no password prompt in
the way.
Pin PATH to trusted system directories once EUID is 0. The restriction lands
only after elevation, so the unprivileged wrapper phase keeps the caller's PATH
and can still find sudo or pkexec; every helper the privileged half uses is a
system utility, so it needs nothing from the checkout.
Clean cherry-pick: both files are byte-identical to quattro, so merging v4-0-1
into quattro resolves without a conflict. Verified by mutation: with the pin
removed, test/shell.d/dns-sudoers-test.sh fails at the poisoned-helper case;
restored, all four of its cases pass. test/shell (189 files) and test/cli pass.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the TOC renumbering (PR #8089 by @samirpokharel, merged to quattro
as b86d4505) onto the v4-0-1 release branch.
Dropping manual/43-extra-themes.md shifted every chapter after it down by one,
on this branch as much as on quattro. The TOC rode along on the old numbers, so
its last ten links pointed at files that are not there: resolving every
manual/*.md link in this branch's README before the change found ten that do not
exist, from 43-extra-themes.md through 52-unattended-installs.md. After it, all
of them resolve.
Clean cherry-pick: the TOC now matches quattro's, so merging v4-0-1 into quattro
resolves without a conflict. test/shell passes: 189 files.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the Taildrop toast fix (PR #7953, merged to quattro as 7e469f96)
onto the v4-0-1 release branch.
A delivery can land hours after it was sent, and the toast announcing it was
expiring after five seconds -- so a file that arrived while nobody was at the
machine was gone from the screen before anyone could click it open. Critical
urgency is what the shell reads as a popup that lives until it is clicked or
dismissed, the same thing omarchy-crash-watch uses to keep its click-to-diagnose
toast around.
The mechanism is already on this branch: omarchy-notification-send parses
options that follow the two positionals, and NotificationLogic.js gives a
critical popup duration 0, which never expires.
One conflict, in test/shell.d/notification-send-test.sh. #7953 added coverage
for the trailing-option path in the notify-send form it had then; #7926, which
merged after it on quattro and is already backported here, rewrote that file for
the Notify D-Bus form and carried the same coverage across. This branch
therefore already asserts what #7953 added -- an urgency and a glyph after the
description reach the call, and the urgency is set once -- so the branch's
version stands and the resolved file is unchanged.
Verified by mutation: with -u critical removed, test/shell.d/tailscale-receive-
test.sh fails at once; restored, it passes, including the new assertion that
every announcement waits to be answered.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the shared URL check (PR #8174, merged to quattro as 68ab12f7) onto
the v4-0-1 release branch, on top of the #8067 backport it follows.
omarchy-theme-install and omarchy-plugin-add both clone a URL a stranger can
choose, and each carried its own copy of the rule that refuses a git option or a
transport helper before cloning. The rule now lives in omarchy-git-url-check and
both callers ask it: two copies of a security check drift, and the second copy
arrived four months after the first only because someone went looking for it.
That rule was also enforcing half of what it described. <helper>::<address> is
one of two shapes git resolves a remote helper from -- it also runs
git-remote-<scheme> for <scheme>://<address> whenever the scheme is not one it
connects itself, so ext::sh -c id and ext://sh -c id reach the same helper while
only the first was refused. Nothing exploitable follows on a stock system:
protocol.ext.allow defaults to never, and the :// spelling hands git-remote-ext
a command name it cannot exec. But a third-party helper installed on PATH is
reachable through the scheme form alone, and a guard is worth more when it
enforces the rule it states.
The :// shape cannot be refused the way :: is, because it is also how every
legitimate URL arrives, so the scheme is checked against the transports git
still connects itself: ssh git git+ssh ssh+git http https ftp ftps file.
git+ssh and ssh+git are on that list because they are spelled like a helper and
read as plain ssh; leaving them off would refuse a URL that clones today. ext
and fd are off it deliberately. A single colon is always scp-style ssh and a
bare path is always a path, so neither needs constraining.
The check fails closed: the callers read a non-zero status as a refusal, so a
missing omarchy-git-url-check refuses the URL rather than waving it through.
Clean cherry-pick on top of the #8067 backport: every file is byte-identical to
quattro, so merging v4-0-1 into quattro resolves without a conflict. test/shell
passes: 189 files, including the one this adds. test/cli passes, so the new
command's metadata is well-formed. Exercised the check here: https, scp-style,
ssh, git+ssh and scp-style IPv6 all pass, while ext:: ext:// fd:: fd:// and a
leading-dash form are refused, as is an uppercase EXT:// spelling.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the sudoless Docker follow-up (PR #8098, merged to quattro as
06a3dbca) onto the v4-0-1 release branch, on top of the #8056 and #8080
backports it follows.
Group membership only takes effect on a fresh session, and in practice a logout
or newgrp is not enough -- only a reboot reliably applies it. The setup and
remove commands now flag the reboot and offer to do it right away with a gum
confirm, the same shape as the GPU toggle, and their notices say "after a
reboot" instead of pointing at logout or newgrp. The existing-user migration
reuses the removal command inside omarchy update, so it passes
OMARCHY_DEFER_REBOOT to skip the prompt there and lets omarchy-update-restart
handle the reboot once the whole update has finished.
Setup > Security showed Sudoless Docker under both Setup and Remove, and the
guards tested the running session's groups, which do not change until the
reboot: after enabling sudoless Docker the menu still offered Setup, the one
action that could no longer do anything, while Remove stayed hidden. Add
omarchy-sudo-docker as the single answer to both questions that differ in that
window -- by default whether this session can reach the socket, which is what
decides if a command must elevate, and with --configured whether the account is
set up for it, which is what the menu and the toggles need. It succeeds when
sudo is needed, so the Setup entry appears while sudoless Docker is off and
Remove once it is on. lazydocker and the Windows VM keep prompting until the
reboot lands.
Clean cherry-pick on top of the earlier backports: every file is byte-identical
to quattro except default/omarchy/omarchy-menu.jsonc, which merged into this
branch's menu and whose two Sudoless Docker lines match quattro exactly.
test/shell passes: 188 files, including the two this adds. test/cli passes, so
the new command's metadata is well-formed. Exercised the helper here: with no
socket it reports sudo is needed, and --configured answers from the account's
groups.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the plugin-add URL guard (PR #8067 by @bastidotnet, merged to
quattro as 30471bf3) onto the v4-0-1 release branch.
omarchy-plugin-add cloned a user-supplied git URL without the transport-helper
guard omarchy-theme-install already applies. That guard arrived with #7884,
which is on this branch, but it never touched plugin-add -- so the sibling
command still leaned entirely on git's own protocol.ext.allow=never to keep a
URL like ext::sh -c <cmd> from running a command at clone time.
Stock systems are unaffected: Omarchy sets no protocol.* override, so the git
default holds and there is no live exploit here. This is defense in depth --
it closes the gap #7884 left in the sibling path and drops a silent dependency
on a default the project does not control. Reject ext::/fd:: and leading-dash
forms; https, ssh, scp-style and token-auth URLs still clone, including an
scp-style IPv6 host, which carries :: of its own.
Clean cherry-pick: both files are byte-identical to quattro, so merging v4-0-1
into quattro resolves without a conflict. The guard reads the same as the one
already in bin/omarchy-theme-install on this branch. test/shell passes: 186
files, with the plugin-add suite's new cases all running here -- including the
pty-driven prompt case, which is the only path that reaches the guard's
leading-dash arm.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Backport of the keypress fix (PR #8082, merged to quattro as 5d3299fb) onto the
v4-0-1 release branch.
gum 2.0 runs a spun command without the terminal attached, so the gum spin --
read -n 1 that held the presentation terminal open returned at once. Every menu
command that ended in a failure took its window down with it before the error
could be read, which is how a failed update looked like a terminal that just
quit.
Read the key directly. gum's own terminal query replies are still sitting on the
tty when the spinner stops, so drain those first or they answer the prompt on
the user's behalf. The /dev/tty node is there whether or not a terminal is
behind it, so open it rather than test for it -- the existence check passed on a
headless run and left both reads failing with "No such device or address" -- and
prompt on the terminal rather than stdout, so a caller that redirects us does
not leave the user waiting on a prompt they were never shown.
The green dot reads better than the globe did, so the provisioning notice uses
it too and drops its spinner along the way.
Clean cherry-pick: both files are byte-identical to quattro, so merging v4-0-1
into quattro resolves without a conflict. Exercised against the installed gum
2.0.0: headless it exits 0 at once, and on a pty it prints the prompt and
returns on one keypress. test/shell (186 files) and test/cli pass.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018DEMYa9UWtroz93DhMTtcV
Spencer Bull