quickshell - Quickshell components for a Hyprland desktop

Age	Commit message (Collapse)	Author	Files	Lines
10 hours	docs: add the notification renderers implementation plan	Danilo M.	1	-0/+1380
	Plan two of two: the quickshell side of the notification daemon. The Notify singleton reads the daemon's published files, a new notifications/ component draws balloons and owns suppression, the drawer's reserved Item becomes the live queue with a history page, and the Status page gains a snooze row. Suppression lives in the balloon shell so the drawer can list what DND held back, and the drawer flag is the single no-double-show mechanism. Per-id history removal is a tracked TODO in the notifyd repo rather than an untracked spec gap.
10 hours	docs: fix the expire_timeout direction in the spec and plan	Danilo M.	2	-20/+27
	The freedesktop spec says -1 means the server decides and 0 means never; the spec and plan had them reversed, which the shipped code inherited and the live handover test caught. Corrected to match the spec, libnotify's -1 default and dunst.
10 hours	docs: correct the snooze check count to five	Danilo M.	1	-1/+1
	The check has five assertions, not four.
10 hours	docs: fix the runtime dir collision in the check script	Danilo M.	1	-1/+2
	The daemon binary was built at $tmp/notifyd, the same name RuntimeDir() gives under XDG_RUNTIME_DIR, so the two collided. The runtime dir moves to $tmp/run, which is what the shipped script does.
10 hours	docs: give the plan's test files the GPLv2 header	Danilo M.	1	-0/+44
	The test snippets omitted the header the global constraints require on every .go file, which a task review flagged against the first one shipped. The snippets now carry it, so later tasks do not repeat the gap.
10 hours	docs: record the notifyd repo as created	Danilo M.	1	-4/+4
	Public under the Linux cgit section, cloned to ~/Programming/GIT/notifyd, so Task 1 Step 1 is done and the executor starts at Step 2.
10 hours	docs: add the notifyd implementation plan	Danilo M.	1	-0/+1780
	Plan one of two for the notification daemon: the Go daemon and notifyctl in their own repo. The renderer plan is separate because the daemon is working, testable software on its own, exercised through notifyctl. The policy is pure and tested without a bus; the store's expiry, dismissal and eviction are separate lifetimes, which the store tests pin; and test-notifyctl.sh runs the daemon and CLI on a private session bus against a temporary runtime directory.
10 hours	docs: separate a notification's balloon and drawer lifetimes	Danilo M.	1	-24/+57
	The freedesktop expire_timeout is the notification's lifetime, and --wait and -b clients block until the daemon closes it. So the daemon emits NotificationClosed at expiry and frees the client, but keeps the entry in the live queue: the drawer lists it until it is dismissed or evicted, inert because its client is gone. History holds only what was dismissed or evicted. The reserved space is therefore scrollable, since the live queue can hold 20 at once, and a row past its expiry is read-only.
11 hours	docs: add the notification daemon design	Danilo M.	1	-0/+276
	The second of the two specs the status registry named. A Go daemon owns org.freedesktop.Notifications and holds the state; quickshell renders it. The daemon and notifyctl live in a separate repo, the balloon shell and the drawer's notification centre live here. Rendering is split because quickshell has no generic D-Bus module, so something outside QML has to own the bus name. Files are the interface in the direction the renderers read, the same convention the registry set, and notifyctl is the one surface back. DND suppression lives in the balloon shell, not the daemon: the drawer's reserved space lists every live notification including the suppressed ones, because a list the user opened is not an interruption. This supersedes the registry spec's assumption that the daemon would read dnd.
11 hours	fix(status): address final review findings	Danilo M.	1	-3/+5
	F1: after a click the shared Switch writes checked directly, dropping StatusRow's declarative binding, so an external mode change from waybar or statusctl no longer moved the switch. Resync via onValueChanged. F2: the spec claimed hyprctl clients counts idle inhibitors, which is false for a layer-surface inhibitor; replace with the behavioral hypridle marker check. F3: README said the check covers both watch states; it exercises the activated report and the absent-file down report. F4: give breakProc an onExited check so a failed breaktimer.sh verb is visible instead of silent. F5: drop StatusTile's unused required st property and its injection. F6: comment the startup-order limit on dndBeforePresentation.
11 hours	docs(status): document the module and record its traps	Danilo M.	1	-5/+35
	A FileView fires its own fileChanged on setText, so a handler that writes in response to a change loops unless it compares first. IdleInhibitor has no window of its own and does nothing without one, so the singleton is handed the keepalive window by shell.qml. Both were read from the documentation while designing and confirmed while implementing.
12 hours	fix(desktop): keep statusctl quiet when the mode file is absent	Danilo M.	1	-6/+6
	read_mode() failed on a missing mode file with a shell redirection error, 'statusctl: line 56: .../status.dnd: No such file or directory', printed to stderr on every get/toggle before the file existed and polluting waybar's exec stderr. The '2>/dev/null' sat on the 'tr' command, but the error is the parent shell's input redirection failure, which that redirect cannot suppress. Guard on '[[ -e "$file" ]]' before reading; a missing file still reads as 0, which was always the contract. Mirror the fix into the plan so plan and script agree.
12 hours	docs: correct the check count and mirror the statusctl hardening	Danilo M.	1	-4/+13
	Two corrections to the plan, both found by running it rather than reading it. The test script the plan specifies makes nine assertions; the prose around it claimed eight, in both the task step and the final check. An implementer running it would see a passing suite that disagrees with its own expected output, which reads as a failure of either the test or the plan. My arithmetic. The plan's statusctl also swallowed a failed rename and read the mode file through cat piped into tr, so a write that did not land still reported success and an unreadable file fell back to off by accident of pipeline semantics rather than by intent. Review of the implemented script caught both. The plan now carries the same fixes, so a future run of it does not reintroduce them. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A7ThHHh5iTYbVfp3rNAkw2
13 hours	docs: set atomicWrites explicitly in the plan's ModeFile	Danilo M.	1	-1/+11
	Review of the first task caught that the commit message credited atomicWrites while the code never set it. The property is already the documented default in 0.3.1, so the behaviour was right, but statusctl watches close_write,moved_to specifically because an atomic write arrives as a rename, and leaning on a default for something the CLI depends on is weaker than declaring it. Both ModeFile blocks in the plan now set it, so task four does not reintroduce the gap when it adds the second mode. Worth recording: the review asserted atomicWrites defaults to false in 0.3.1 and concluded the shipped writes were not atomic. The v0.3.1 and v0.3.0 documentation both read "If true (default)", so that conclusion was wrong, though the underlying observation about the unset property was right. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A7ThHHh5iTYbVfp3rNAkw2
13 hours	docs: add the status registry implementation plan	Danilo M.	1	-0/+1216
	Eight tasks against the registry spec: the singleton with the dnd mode, the statusctl CLI and its check, naming the keepalive window, presentation mode and its three effects, the module and tile, the page and registration, installing the CLI and swapping the waybar module, then the README and traps. The plan carries the facts probed while designing so the implementer does not re-probe them: the runtime directory is tmpfs, elogind runs here with the pam caveat that stops it guaranteeing logout cleanup, the compositor advertises zwp_idle_inhibit_manager_v1, inotifywait is present, and Inconsolata Nerd Font lives under ~/.fonts rather than either share directory. That last one cost a failed cmap search that read as a missing glyph when it was a bad search path. Two claims stay marked unverified because they come from documentation and have not been observed running: that a FileView fires its own fileChanged on setText, and that IdleInhibitor does nothing without a non-null window. The plan handles both and tells the implementer to confirm them and record what actually happens. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A7ThHHh5iTYbVfp3rNAkw2
13 hours	docs: add the status registry design	Danilo M.	1	-0/+229
	Desktop modes as state: do not disturb and presentation mode, with one place that owns them. First of two specs; the notification daemon that replaces dunst is the second and consumes the registry's dnd mode rather than owning it. The registry comes first because DND belongs to the desktop rather than to the notification daemon, and because the daemon is the larger and riskier component. The design that needed deciding was not the toggle but the coupling. Presentation mode sets DND, so DND has two writers, and turning presentation mode off has to restore what the user had rather than clearing it. That rule is why this is a registry and not two booleans. State is one file per mode in XDG_RUNTIME_DIR, matching the convention breaktimer already writes on this machine. That directory is /run/user/1000, a tmpfs, so a reboot clears every mode with no cleanup code. Logout without reboot is less certain and the design does not lean on it: elogind runs here and its pam module documents removing the runtime directory at last logout, but the same manual says the module does nothing when the system was not booted with elogind as init, which on Slackware it is not. If removal does not happen, a mode survives a logout, which is the same behaviour as surviving a shell restart. FileView covers both directions: atomicWrites for the write, watchChanges for the watch, no shell-out. External watchers need close_write,moved_to because an atomic write arrives as a rename, the same trap the mail watcher hit with Xapian. Confirmed while designing: the compositor advertises zwp_idle_inhibit_manager_v1 and waybar's built-in idle_inhibitor already drives it here, so the inhibitor is asserted over Wayland rather than D-Bus. That module is replaced rather than demoted, because it owns its inhibitor object and cannot display state owned by anything else; left alongside the registry it would assert a second inhibitor and idle would resume only when both released. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A7ThHHh5iTYbVfp3rNAkw2
29 hours	docs: KDE Connect module design and plan	Danilo M.	2	-2/+1226
	Absorbs the kdeconnect-indicator into the drawer: status, battery, pairing, ping, clipboard send, file share, refresh and mount. State is polled from a qdbus6 helper script while the drawer is open, since quickshell 0.3.1 has no generic D-Bus module. The spec notes the phone-to-PC clipboard is the daemon's plugin, not the indicator's, so removing the indicator does not touch it.
29 hours	docs: KDE Connect state protocol carries the pair flags	Danilo M.	1	-8/+14
	The pairing flow needs isPairRequested, isPairRequestedByPeer and the verification key, so the polled line carries them; the key is only queried with a pairing in flight. A failed daemon query is a non-zero exit with no output, which is how the module tells it apart from a valid empty list. Pairing failure comes from the kdeconnect-cli exit code, not the pairingFailed signal, since there is no generic D-Bus module to watch it.
29 hours	docs: KDE Connect module design	Danilo M.	1	-0/+217
	Status plus basic actions, absorbed from the indicator. State comes from qdbus6 over kdeconnectd, since quickshell 0.3.1 has no generic DBus module. Polled while the drawer is open, driven by the tile lifecycle. Notes the indicator is not what serves the phone-to-PC clipboard, and the FileDialog focus risk under an Exclusive layer surface.
30 hours	docs: the network tile is a globe and reads the local IP	Danilo M.	1	-4/+13
	The tile no longer swaps a wired/wifi icon, and the local IP comes from ip -j because Quickshell.Networking exposes only the MAC. Recorded in the spec so it matches the code.
34 hours	docs: correct three plan details found during execution	Danilo M.	1	-14/+32
	The Connections block needs a typed property on a QtObject, the bluetooth README lines belong to Task 6 so every commit stays self-consistent, and Pairing has no status property. Each was found by a task review and fixed in the plan as well as the code.
34 hours	docs: plan the network and bluetooth modules	Danilo M.	2	-5/+1350
	Eight tasks, each leaving the tree working: network wired page, wifi section, README, bluetooth module and page, pairing, README, then the waybar removal and the trap notes. The spec's failure section is corrected in the same commit: it claimed forget and disconnect failures would surface inline, but neither backend exposes a failure signal for them, so that is a ceiling rather than a feature. Verified by probe before writing any task: the singletons start empty for about two seconds, the device lists are ObjectModels, signalStrength is 0..1, scanner and discovering are writable, and NMSettings is not in QML scope, which drops the hidden-network join the spec had left conditional.
34 hours	docs: design for network and bluetooth modules	Danilo M.	1	-0/+160
	Two drawer modules covering wired, wifi and bluetooth, using the native Quickshell.Networking and Quickshell.Bluetooth backends rather than nmcli or bluetoothctl. The one exception is pairing: quickshell ships no BlueZ agent and no generic D-Bus module, so a device needing passkey confirmation is paired through bluetoothctl, whose own agent handles it. The wifi module became Network so a connected eth0 does not mask the wifi state; the tile shows both links when both are up because the API exposes no route metric. Replaces waybar's network and bluetooth indicators, per the drawer being the single place they are managed.
35 hours	docs: correct the plan's no-arg ipc open and the head pipe	Danilo M.	1	-6/+20
	Quickshell 0.3.1 IPC requires every declared argument, so the plan's `ipc call drawer open` with no page fails; the page-less grid entry is the zero-argument `toggle`. Also records that `... \| head` races timeout and kills the shell early, so verification uses a file redirect.
37 hours	docs: a subdirectory singleton needs no qmldir, drop it from the plan	Danilo M.	1	-71/+54
	The plan claimed a pragma Singleton outside the config root is invisible until a qmldir names it, and had Task 1 create one plus Tasks 6, 7 and 8 append a line each. That premise was wrong. Tested: a pragma Singleton in modules/sub/, reached by a plain import "modules/sub", resolves with no qmldir anywhere. The control was a reference to a genuinely undefined type, which warns "ReferenceError: <name> is not defined"; the singleton case produced no such warning and the binding evaluated. This matches what AGENTS.md already documented about Theme.qml, which the code review pointed out. Four registration steps removed and the remaining steps renumbered. The AGENTS.md note Task 9 adds now records the tested behaviour rather than the invented one. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01We9zcHEH8ZWHPcQwMaJRNv
37 hours	docs: warn that the config task's tilde paths must be typed absolute	Danilo M.	1	-0/+9
	The paths are written with a tilde because this file is committed and the repo forbids home paths in committed files. But neither Hyprland's exec_cmd nor waybar's exec goes through a shell, so a tilde there never expands and fails silently, which reads as a broken keybind rather than a bad path. The two rules genuinely conflict here, so the plan now says which form goes where instead of shipping lines that would not run. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01We9zcHEH8ZWHPcQwMaJRNv
37 hours	docs: implementation plan for the desktop shell	Danilo M.	1	-0/+2817
	Ten tasks: skeleton, contract, chrome, drawer, then the four modules in increasing order of risk, then documentation, then the six config edits outside this repo. Appearance goes first among the modules despite being last in the grid, because it is nine lines of substance and puts something in the grid before the three migrations start. Sound, mail and vm follow in that order: sound splits one 302-line file into three, mail is the gentlest move and carries the only automated oracle in the project, and vm is the largest and the one whose polling behaviour actually changes. Each task leaves the repo working, because several delete a component and a half-finished deletion is painful to unpick. Paths in the config task are written with a tilde: the gitleaks hook rejected the absolute form, correctly, since this file is committed and the live Hyprland config it describes is not. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01We9zcHEH8ZWHPcQwMaJRNv
37 hours	docs: settle the vm tile's cold state on drawer open	Danilo M.	1	-3/+4
	The vm module's service is lazy, so the tile had no state to show before its page had been visited. One `virsh list` when the drawer opens fixes that for the cost of a single call; the per-VM polling for guest-agent stats and managed-save detection still waits for the page, which is where the expense actually is. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01We9zcHEH8ZWHPcQwMaJRNv
37 hours	docs: design for a unified desktop shell with a drawer and modules	Danilo M.	1	-0/+271
	Five components mean five processes, five autostart lines, five blur rules, and functionality scattered across surfaces with nothing in common. Nothing is discoverable without remembering it exists. The design is one left-side drawer on DP-1, reached from a launcher at the left end of waybar, hosting modules as tiles that may open full-height pages. Modules declare a tile, a page, both or neither, plus whether their background service runs while the drawer is closed: sound and mail must (the OSD reacts to keypresses, the unread count outlives the drawer), vm must not, since polling virsh only paints a page nobody is looking at. Three components merge in (volume-osd, mail-overview, vm-manager), two stay separate because a 600px drawer is the wrong shape for them: window-switcher is a fullscreen ALT+TAB grid, appearance needs room for a wallpaper picker. Appearance still earns a tile that fires its existing IPC, which is the case that proves a tile needs no page of its own. The top of the drawer is reserved and empty. The notification engine is a later project, but the space has to be claimed now: build the grid anywhere else and it moves the day notifications land, having been tuned against a position that does not survive. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01We9zcHEH8ZWHPcQwMaJRNv
2 days	fix(mail-overview): don't re-notify the prior batch, flatten newlines	Danilo M.	1	-5/+10
	The lastmod range is inclusive at both ends and the stored revision was reused as the next start, so the batch just notified matched again. Start one revision later. A subject can contain a newline, which broke the bullet rows and the +N more count; flatten CR/LF to a space before escaping. Also reconciles the state-write comment and spec with the actual skip-and-advance failure behavior, and corrects two doc wordings.
2 days	docs(mail-overview): document the arrival notifier	Danilo M.	1	-20/+26
	Why a revision counter rather than a count delta or a date watermark, why the state file carries a UUID, and why an unseeded run stays quiet. Also records the final notification presentation: the split heading, the absolute-path icon, and the show_indicators setting that keeps the action indicator off.
2 days	docs: implementation plan for mail arrival notifications	Danilo M.	1	-0/+924
	Six tasks, TDD against one bash check rather than a framework: the script is sourced as a library so the pure functions, the ones that only move text around, are asserted without notmuch, dunstify or inotify in the picture. The two parsing traps that already cost this component a debugging session each get a fixture apiece: a key containing dots, and a folder value containing a bracket. Verification does not wait for mail. Backdating the stored revision by 2000 replays a real window on demand, and the seed run before it proves the anti-storm rule with 101 unread messages on disk. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017ZwWrCEbdKzmisfg9bb1nS
2 days	docs: design for mail arrival notifications	Danilo M.	1	-0/+183
	One notification per account per arriving batch, from a new mail-notify.sh sibling to waybar-mail.sh. Arrival is detected with notmuch's lastmod revision counter rather than a count delta or a date watermark: a commit is not the same as new mail, since reading and tagging also commit, and date: is the message's own Date header, so backdated mail would never notify and future-dated mail would notify forever. A separate process rather than extending waybar-mail.sh, which already has the arrival edge: waybar owns that process, so a bar restart would stop notifications with nothing reporting it. State carries the database UUID beside the revision, because notmuch revisions are only comparable within one database. Missing, corrupt or mismatched state seeds silently: with no floor, lastmod:0.. matches all 101 unread messages here and startup becomes a wall of popups. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017ZwWrCEbdKzmisfg9bb1nS
3 days	docs(window-switcher): focusHistoryID is not on HyprlandToplevel	Danilo M.	1	-2/+12
	Implementing the model found it. The property reads undefined, so the planned sort compared NaN and silently did nothing, leaving the list in arbitrary tracker order while looking entirely plausible. It is on lastIpcObject instead. Carried into the plan's later tasks so the grid does not reintroduce the bare property, and into the notes that Task 6 puts in AGENTS.md, since a sort that quietly does nothing is the kind of wrong answer this repo's notes exist to prevent. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYg4wYHq5XNbiVmMeKRb1S
3 days	fix(window-switcher): pick the screen the way the other components do	Danilo M.	1	-1/+4
	The screen was found by a name written inline, falling back to null. The other three panels all take a monitor property and fall back to screens[0], so this now does too: a machine without a DP-1 gets its first monitor rather than a null screen, and the name is in one place if it ever needs changing. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYg4wYHq5XNbiVmMeKRb1S
3 days	docs(window-switcher): the implementation plan	Danilo M.	1	-0/+849
	Six tasks, each ending in a commit: the skeleton and the keyboard grab, the window model, one card, the grid and empty state, the Hyprland wiring, and the docs. Two things were checked against the running system while writing it rather than left for the implementation to discover. The keyboard grab works: Keys.onEscapePressed on a focused Item inside the layer surface fires, so the AGENTS.md workaround holds and keyboard navigation is safe to build on. And the dispatcher for sending a key is send_shortcut with the underscore; sendshortcut does not exist, and a test written against that name would have failed in a way that looks exactly like the key never arriving. No qmldir is added. There is none anywhere in this repo and Theme resolves without one, since quickshell scans the config directory itself; the AGENTS.md note about needing one is about singletons outside it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYg4wYHq5XNbiVmMeKRb1S
3 days	docs(window-switcher): an empty state, rather than opening on nothing	Danilo M.	1	-0/+19
	ALT + TAB with no windows would otherwise give a full-screen dim with nothing in it, which reads as a hang. It is reachable from a fresh session, and from closing the last window in the switcher itself, which stays open by design. It now says so instead. The empty state has no card to click, so the backdrop closes the overlay on click as well as on Escape. That applies generally, not only when there is nothing to show. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYg4wYHq5XNbiVmMeKRb1S
3 days	docs(window-switcher): dispatch args are Lua, and focus raises in monocle	Danilo M.	1	-6/+31
	Two findings from testing against the running compositor, both of which would have cost time during implementation. Hyprland 0.56.2 evaluates dispatch arguments as Lua, so the form this design previously specified, dispatch focuswindow address:0x..., is a syntax error rather than a command. It fails silently unless stderr is read: the window does not focus and nothing is reported. hypr-windows.sh was written with hl.dsp.focus({ window = "..." }) for that reason, and this design had misread that as legacy oddity. The fullscreen batch sequence from that script is not carried over. Workspaces 1 and 8 are monocle and 4 and 5 are scrolling, so a stacked window is the normal case here rather than an edge case, and the worry was that focus would land without the window becoming visible. Tested with two stacked windows on workspace 1: focus alone raised the buried one, confirmed on screen in both directions, so no alterzorder and no batch sequence are needed. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYg4wYHq5XNbiVmMeKRb1S
3 days	docs(window-switcher): cards fit both orientations, and grow to a ceiling	Danilo M.	1	-7/+32
	The card cannot take its shape from its preview. DP-3 reports transform=1, so its windows are 1076x1916 portrait against 2556x1034 on DP-1, a 4.4x span of aspect ratio in one grid, and a fullscreen app on that monitor is portrait too. So the card is a fixed 16:10 box with the preview fitted inside it, which keeps rows aligned; sizing each card to its own preview would give ragged rows and break the alignment of the three text lines that makes the grid scannable. Cards now grow to fill their row up to 640px rather than sitting at a fixed size, so a lone window is shown large without a preview becoming a full-screen mirror of the window it stands for. Column count is the window count capped at six, centred, so one card sits dead centre and two straddle the middle. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYg4wYHq5XNbiVmMeKRb1S
3 days	docs: design for the window-switcher component	Danilo M.	1	-0/+159
	A full-screen grid of live window previews to replace the rofi list on ALT + TAB. The script it replaces is not broken, so this is an upgrade from a textual list to something that shows what is in each window. Settled during design: toggle-style rather than hold-and-release, since a missed ALT release would leave a stuck full-screen overlay; cards wrap into a grid rather than shrinking, so a high window count degrades by growing downwards; only special: workspaces are filtered, because jumping to another desktop is the point of a switcher. Feasibility was checked against quickshell 0.3.1 rather than assumed. ScreencopyView is compiled in and captured all four open windows, the scratchpad on special:special included, so hidden windows need no icon-only fallback. Hyprland.toplevels reads 0 until refreshToplevels() is called, and its address field omits the 0x that dispatches need. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYg4wYHq5XNbiVmMeKRb1S
4 days	docs: the plan does not launch anything on the user's screen	Danilo M.	1	-4/+11
	The user works at this machine and runs the shell themselves. An agent executing this plan writes code and runs non-GUI checks, then hands over the command and the numbers to compare against; the four steps that need a running qs are now marked as the user's. Also warns against a bare pkill -x qs, which would take down the three components the user already has running. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WWL8JYHu7yhAdtx5pU9PMU
4 days	docs: correct the notmuch malformed-query claim	Danilo M.	2	-11/+23
	The previous commit, the spec and the plan all said notmuch exits 0 on a malformed query while printing something that is not a count, and that validating the output as an integer therefore catches it. Measured properly, that is wrong in a way worth recording, because the truth is worse. notmuch fails two different ways. A rejected query prints nothing and exits 1: `notmuch count 'tag:unread and ('`. A query Xapian merely misparses returns a plausible wrong number and exits 0: `notmuch count 'tag:unread and (('` gives 41, and `'tag:unread and tag:'` gives 3. The second is undetectable by any check on the output, which is why the original claim was not just imprecise but inverted: the case it described as caught is the case nothing can catch. The integer validation still earns its place, on the first failure mode, where empty output would otherwise render as an empty inbox. The real defence against the second is that QUERY is a fixed string and is never built from anything, which the comment now says. The earlier measurement that produced the wrong claim read 40 as mangled output when it was a successful parse answering a different question. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WWL8JYHu7yhAdtx5pU9PMU
4 days	docs: implementation plan for mail-overview	Danilo M.	2	-13/+1363
	Seven tasks, ordered so the thing most likely to be wrong is checked first. The watcher script comes before any QML because it is the only piece with a real runnable check and needs no compositor. The component skeleton gets its own commit because the failure it guards against is invisible: a config whose only window is hidden exits straight after logging Configuration Loaded and reports no error. Two decisions changed while reading the existing components. The drawer is a fullscreen overlay with content anchored top right, which is the idiom both other panels use and what gives click-outside and Escape for free, rather than a small anchored window. And its exclusionMode is Normal rather than the Ignore those two use, so waybar's exclusive zone places it below the bar without this component carrying the bar's 42px height as a constant to drift. Also records a trap measured this session: notmuch exits 0 even for a malformed query, printing something that is not a count, so every count is validated as an integer rather than trusting the exit status. A failure that rendered as 0 would look exactly like an empty inbox. The live-config task needs real absolute paths, because waybar's exec and Hyprland's Lua strings do not expand a tilde. They are written as <repo> and <home> placeholders with a note to substitute, since the gitleaks hook blocks committed home paths and was right to reject the first attempt. The spec was updated to match both changes. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WWL8JYHu7yhAdtx5pU9PMU
4 days	docs: record why mail-overview counts by tag, not path	Danilo M.	1	-0/+23
	qtmaildir scopes an account with a path glob, so the obvious move was to copy that query. It is wrong for this panel. notmuch deduplicates by message id, so a message that arrived at two configured addresses is one message with two paths, and a path glob counts it under both accounts: the five accounts here sum to 102 against a global total of 101. The icon shows one global total and the panel header repeats it, so rows that sum to a different number would look like a bug. Counting by the account tag, which is a property of the message and therefore singular, makes the rows sum to the header. The cost is that a cross posted message appears under only the account the post-new hook attributed it to. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WWL8JYHu7yhAdtx5pU9PMU
4 days	docs: design for mail-overview, a notmuch mail drawer	Danilo M.	1	-0/+259
	Replaces three waybar modules that each poll the Gmail API through a python script with its own credentials file, cover three of five accounts, and open Thunderbird. Everything they fetch over the network is already in the local notmuch index, tagged per account. Two decisions carry most of the design. Counts are scoped to "tag:unread and tag:inbox" rather than plain "tag:unread", which for one account is the difference between 32 and 41, because the inbox number is the one that means new mail worth looking at. And the waybar module runs in continuous mode with an inotifywait loop, so waybar owns the watcher process and there is no daemon to supervise on a machine with no systemd; the watch is on the xapian directory rather than on named files because a commit replaces them, which breaks a watch held on a file. The account list is not in the component. qtmaildir.conf already has one [account.<key>] section per account, where <key> is the notmuch tag suffix, carrying a display label and a colour, so the panel parses that and a new account appears without touching QML. Thread level actions are out of scope because they are blocked rather than deferred: qtmaildir takes no arguments, so nothing can tell it which thread to open. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WWL8JYHu7yhAdtx5pU9PMU