#+TITLE: ArchSetup Tasks
#+AUTHOR: Craig Jennings
#+DATE: 2026-02-14


* Archsetup Priority Scheme
Four levels, matching the Emacs config (=org-highest-priority ?A=, =org-lowest-priority ?D=, =org-default-priority ?D= in =modules/org-config.el=). Priority answers "how much does this matter"; a date answers "when". They are independent — assign both deliberately. Org priority alone never schedules anything, which is why undated [#A]/[#B] tasks feel ungrounded.

- [#A] Must happen. Broken install, data loss, security, or a blocker for other work. An [#A] REQUIRES a SCHEDULED or DEADLINE date — if it can't be dated, it isn't really an A; drop it to B. (The main agenda always shows open A's.)
- [#B] Should happen, this cycle. Real improvement or fix with no hard date. Surfaces in the agenda's priority-B block only while undated; add a SCHEDULED date when you commit to a week and it moves into the schedule.
- [#C] Nice to have / someday. Kept for the record, low urgency. Date it only when it graduates to B.
- [#D] Default / unsorted. A bare TODO with no cookie is D. Stays out of the agenda — the inbox of priorities. Triage D's up to A/B/C or let them sit.

Rule of thumb: A = dated-and-must; B = the active backlog; C = parking lot; D = untriaged. Fixing the undated A/B tasks means either dating them or demoting to C.

** Tags

The vocabulary is open — topic tags are coined as needed — so these are conventions, not a closed set. A task carries at most one type tag, optionally the effort/autonomy tags, and any number of topic tags. Because the set is open, the task audit leaves topic tags alone (it doesn't strip "unknown" tags).

- *Type* (one per task where the kind is clear): =:feature:= new capability, =:bug:= fix for broken behavior, =:test:= test coverage or test infra, =:refactor:= restructure with no behavior change, =:chore:= tooling / meta / housekeeping.
- *Effort / autonomy*: =:quick:= a spare-moment fix (minutes, not a sitting); =:solo:= Claude can carry it end to end — there's a build path, a test path, and no upfront decision needed (a leftover manual spot-check doesn't disqualify it).
- *Topic / area* (open): the subsystem a task touches — e.g. =:hyprland:= =:waybar:= =:mpd:= =:music:= =:network:= =:tooling:= =:llm:= =:eask:= =:pocketbook:= =:cmail:=. Coin a new one when it aids filtering.
* Archsetup Open Work
** TODO [#A] Velox boot recovery — no kernel in BE :bug:velox:zfs:
SCHEDULED: <2026-07-16>
Velox won't boot: ZBM prompts for the passphrase, unlocks, then reports no bootable environment with a kernel. Cause: an interrupted kernel =-Syu= removed the old kernel and never installed the new one — /mnt/be/boot (from zroot/ROOT/default) holds ONLY intel-ucode.img; vmlinuz-linux + both initramfs are gone. /boot lives inside zroot/ROOT/default (no separate boot dataset), so root-dataset snapshots capture it.

Status 2026-07-15: a first rollback attempt did NOT fix it (square zero after reboot) — suspected typo in the snapshot name, so the rollback likely errored and did nothing. NOT verified. Next session: verify state in the ZBM recovery shell BEFORE any reboot.

Recovery lever: the pre-pacman ZFS snapshot hook (live on velox since 2026-06-29) snapshots zroot/ROOT/default@pre-pacman_<ts> before every pacman transaction. The newest =pre-pacman_<ts>= predating the failed upgrade holds the intact old kernel — roll back to it.

Morning steps (Craig at velox ZBM → recovery shell, Ctrl+R):
#+begin_src sh
# 1. pool writable + key loaded
zpool get readonly zroot
zfs get -H -o value keystatus zroot/ROOT/default
#   if readonly=on: zpool export zroot && zpool import -f -N zroot
#   if keystatus=unavailable: zfs load-key zroot

# 2. list snapshots — COPY THE EXACT NAME (the typo bit here last time)
zfs list -t snapshot -o name,creation zroot/ROOT/default | grep pre-pacman

# 3. see current /boot state (read-only mount)
umount /mnt/be 2>/dev/null; mkdir -p /mnt/be
mount -t zfs -o zfsutil,ro zroot/ROOT/default /mnt/be
ls -la /mnt/be/boot

# 4. if /boot still shows only intel-ucode.img: redo rollback with the exact name
umount /mnt/be 2>/dev/null
zfs rollback -r zroot/ROOT/default@pre-pacman_<EXACT-TS>   # -r, NOT -R

# 5. VERIFY before reboot — remount RO, confirm the kernel is back
mount -t zfs -o zfsutil,ro zroot/ROOT/default /mnt/be
ls -la /mnt/be/boot   # MUST show vmlinuz-linux + initramfs-linux.img
umount /mnt/be

# 6. only once /boot shows a kernel:
zpool export zroot && reboot
#+end_src
Scope: only zroot/ROOT/default reverts; /home, /var, /media are separate datasets, untouched. After boot: =pacman -Syu= attended, confirm /boot holds vmlinuz-linux + initramfs before any shutdown. Full diagnosis: =inbox/PROCESSED-2026-07-15-0002-from-.emacs.d-velox-boot-failure-handoff.org=; ZBM photo: =inbox/PROCESSED-2026-07-15-0002-from-.emacs.d-PXL_20260715_043758976.jpg= (local on ratio; inbox is gitignored).

** TODO [#B] Velox boot-failure retrospective — upgrade guard gaps :bug:zfs:maint:
Post-mortem for the 2026-07-15 velox no-kernel boot failure, from the archsetup/maint code review:
- maint's UPDATE remedy runs a plain =yay -Syu --noconfirm= (remedies.py:297). The live-update guard (guard.py) only matches mesa/hyprland (the 2026-06-07 live-swap class) — it never checks /boot, kernel, initramfs, or mkinitcpio exit. No post-upgrade /boot assertion exists. An interrupted kernel transaction slips straight through.
- Add a post-upgrade /boot assertion: after a transaction touching linux/linux-*, confirm vmlinuz-* + initramfs-*.img present and mkinitcpio exit 0; refuse to end the run (or page Craig) otherwise. Would have caught this.
- Sanoid-vs-actual dataset drift: configure_zfs_snapshots configures zroot/var/log + zroot/var/lib/pacman as separate datasets; velox's actual layout has neither separate (/var/log sits inside zroot/var). Reconcile.
- Confirm the pre-pacman snapshot hook is actually installed + firing on velox (it should be — it's what makes recovery possible).

** TODO [#C] Add inetutils to install base :feature:solo:quick:network:
TRAMP's /ftp: method (ange-ftp) shells out to a command-line ftp client; Arch ships none by default. GNU inetutils provides =/usr/bin/ftp=. Craig's dirvish config has an FTP quick-access entry (phone FTP server), so it's a config dependency. Installed manually on ratio 2026-07-14; velox needs it once it boots. Add to the install base so future machines get it for free; verify via VM test. From .emacs.d handoff 2026-07-14-1751.

** DOING [#B] Widget gallery upgrades :feature:design:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-13
:END:
Usability + documentation pass over the [[file:docs/prototypes/panel-widget-gallery.html][panel widget gallery]], orthogonal to the component-generation spec work, so it runs on the =gallery-upgrades= branch (squash merge to main after Craig's UI confirmation + tweaks). Items 1-4 run as a no-approvals speedrun (Craig authorized 2026-07-12); item 5 is a joint brainstorm after the merge.
*** 2026-07-12 Sun @ 12:59:48 -0500 Added the card size toggle (1x/2x/3x, default 3x)
Masthead size chips apply CSS zoom per grid; 2x/3x drop the 1320px wrap cap so wide monitors get the room. CDP-verified: 84 cards, no exceptions, fader drag and toggle click both track at 3x (drag helpers are rect-ratio based, so zoom is transparent to them).
*** 2026-07-12 Sun @ 13:42:29 -0500 Retuned the scale split per Craig
Craig: cards a fifth smaller, fonts and palette back to normal but a little larger. The toggle (now S/M/L, "widget size") zooms only the stage — L is 2.4x, M 1.7x — while card text sits at its own raised base size (wname .95rem, note .85, spec-sheet .82) and the palette left the zoom rules entirely (tiles bumped 148→176px, fonts up a notch). Grid columns widen per size step instead of zooming. Probes re-run green.
*** 2026-07-12 Sun @ 13:52:11 -0500 Raised all reading text to 12-13pt per Craig
One move: html font-size 130%, scaling every rem-based size (notes/spec sheets/palette/masthead/readouts). Verified computed: notes 13.3pt, spec sheets 12.8pt, palette 12.5pt — was ~10pt. Widget-internal px sizes untouched (they are drawing, scaled by the stage zoom).
*** 2026-07-12 Sun @ 14:07:00 -0500 Added R32 mechanical timer dial from Craig's reference
Wind-up interval timer (Controls, after R29; gallery at 85 cards): knurled coin-edge dial rotating under a fixed red index, OFF/20/40/60 at 4.5°/min, hub screw, winding arrow, PUSH TO STOP / TURN TO START engraving, red fluted stop knob. Drag winds (turn to start), the dial runs itself back down at a demo minute-per-second (reduced-motion gated), red knob click zeroes it (push to stop). Spec sheet included. Reference filed (2026-07-12-mechanical-timer-dial.png). CDP-verified: wind → T-35 MIN, tick 35→34, stop → OFF, zero exceptions.
*** 2026-07-12 Sun @ 14:21:27 -0500 Added R33 four-way rocker from Craig's reference
Navigation pad (Controls, after R32; gallery at 86 cards): recessed plate, rubber pad, four outward silver arrows, center pivot nub. Quadrant-sized click zones step a tracked cursor (readout shows direction + x/y), pressed arrow flashes gold. Click-only — noted in the spec sheet as the most Emacs-portable control in the kit (it is literally arrow keys). Reference filed (2026-07-12-four-way-rocker.png). CDP-verified: all four quadrants step and accumulate correctly, zero exceptions.
*** 2026-07-12 Sun @ 14:26:20 -0500 Timing-marker dial judged covered (Craig concurred: probable dupe)
Scope timing-marker reference (VARIABLE TIMING / MARKER SEC): a knob rotates a printed scale disc under a fixed top index — that mechanism is R02 (vernier disc under hairline), and the value-in-a-window read is R20 (drum roller). BANKED enrichment, new to the kit: the backlit active-value window — the full scale stays visible but only the selected value glows through an amber window (R02 shows all + highlights none; R20 highlights one + hides the rest). Also noted: the 1-2-5 stepped decade scale (us/ms/s) as detent content for any stepped dial. Reference filed (2026-07-12-timing-marker-dial.png); no new card.
*** 2026-07-12 Sun @ 14:29:22 -0500 Added R34 four-way toggle selector from Craig's reference
NOT a dupe of R33: the rocker is momentary (press = step), this is stateful (the lever lives in A/B/C/D). Test-panel styling: printed X/Y axes + circle + diagonal legend square, A-D label rings, corner lamps wired by printed lines, chrome ball lever that throws to the clicked quadrant while its lamp takes the light (exactly one lit, jewel-a). Click-only. Gallery at 87 cards. Reference filed (2026-07-12-four-way-toggle.png). CDP-verified: default POS C per the photo, all quadrants select, single-lamp invariant holds, lever transform tracks, zero exceptions.
*** 2026-07-12 Sun @ 14:37:46 -0500 Added R35 day-date disc calendar from Craig's reference
Watch day-date complication (Indicators, after R31; gallery at 88 cards). Judged NOT covered: N04 is two nested input knobs, R02 one disc under a hairline — a compound READOUT from independent coaxial discs is a new indicator form. Cream disc, 31 radial date numerals (upside down at bottom like the movement), weekday names twice around inside, fixed yellow hand as the read index, movement-plate screws. Live: initializes to today; click rolls midnight one day (both discs step). Spec sheet notes the honest limitation — the 31-slot disc ignores short months, exactly like the hardware. Reference filed (2026-07-12-daydate-discs.png). CDP-verified: today correct (SUN 12), click → MON 13, both disc transforms track, zero exceptions.
*** 2026-07-12 Sun @ 14:40:47 -0500 Added R36 LED dot matrix from Craig's reference (answered: what display is this)
Answered Craig's question: an 8x8 red LED dot-matrix module (64 discrete LEDs behind a tinted window, multiplexed; Kingbright/Lite-On family part) on Kilpatrick Audio's K4816 Pattern Generator, drawing its K logo. Judged new form: seven-seg/starburst are fixed segments, VFD marquee is character cells, R10 is a text LCD — a free bitmap display is none of those. Built as R36 (Indicators; gallery at 89 cards): 8x8 paintable matrix starting on the reference K (17 dots), click toggles any dot, readout counts lit. Reference filed (2026-07-12-led-dot-matrix.png). CDP-verified: K = 17/64, paint on → 18, paint off → 17, zero exceptions.
*** 2026-07-12 Sun @ 18:11:02 -0500 Built the takuzu-survey Tier 1+2: R37-R47 + N23 alarm lifecycle (gallery at 100 cards)
Craig picked option 2 from my assessment of takuzu's historical-panel-components research (60+ candidates; ~a fifth of its "new" verdicts were stale against the current kit, and the magic-eye entry re-litigated a removal takuzu itself requested 2026-07-11). Built + CDP-verified in 4 commits: 9b625d6 R37 pin routing matrix (VCS3 many-to-many, pins seat/pull, route count) + R38 dead-man button (first held-state control: pointer-capture, dwell clock, release-to-safe) + R39 rotary telephone dial (wheel winds to the stop and returns, digit pacing authentic); 109b3dd R40 breaker panel (first system-thrown control: TRIP pops a breaker to the amber mid-state, two-step reset) + R41 DSKY (verb/noun grammar, V35 lamp test, OPR ERR on bad grammar — one bug caught by probe: stale verb register re-executed on bare ENTR, fixed) + R42 cam-timer drum (procedure-as-position, self-advances after a click starts it); eb926ea R43 attitude indicator (first two-axis instrument, 2D drag) + R44 heading bug + servo needle (commanded-vs-actual, shortest-path chase); cbcad5b R45 flip-disc array (bistable pixel, scaleX flip) + R46 dekatron (pulse-stepped glow, carry blink on wrap) + R47 landing gear (spatial three-greens + transit pulse) + N23 upgraded with the alarm lifecycle (flash → ACK steadies → new alarm re-arms → RESET / TEST). All spec sheets written. BANKED as idioms, not cards: mimic diagram, oscilloscope cluster, throttle quadrant, voice-loop strip, blinkenlights, lockout-tag disabled-with-reason, knife-switch skin, decade-switch bank, normalled-jack overlay, neon-vs-jewel, VHF detent, valve handwheel, circular chart, two-hand anti-tie-down. Judged covered/stale in the doc: warning flag (R11), split-flap (N20), strip-chart (N16), odometer (N25), interval timer (R32), telegraph (R30), foot switch (R24), light gun/trackball (the mouse), guarded-toggle two-step (R29 already does it). Full regression green at cbcad5b.
*** 2026-07-12 Sun @ 18:29:52 -0500 Control-grammars reference note + the six promoted banked cards (gallery at 106)
Craig picked option 3 (note first, then build) and asked for the reading list in home's inbox. Shipped: docs/design/2026-07-12-control-grammars-reference.org (b0cebcf — the seven named grammars each with a kit exemplar, the literature from Moran CLG through MIL-STD-1472/NASA HIDH, and seven proposed taxonomy axes; the gap-finding move is crossing task x cardinality for empty cells). Reading list delivered to home/inbox (2026-07-12-1822). Then promoted six banked items to cards, CDP-verified in two commits: 5ef616f R48 knife switch (side-view blade on a real hinge; the air gap is the proof) + R49 decade box (four skirted decade knobs compose 3,500 Ω style values) + R50 two-hand safety (arm window 0.5s, same-hand and timeout both fault); 8b4978b R51 voice-loop keyset (independent monitors, exclusive talk, per-loop activity flicker) + R52 blinkenlights (live ADDR/DATA lamps folding in a clickable switch register, octal readout) + R53 circular chart recorder (day-per-revolution pen trace, fresh paper on click). Full regression green at 8b4978b, 106 cards. Still banked (genuinely idioms): mimic diagram, scope cluster, throttle quadrant, lockout tag, knife-skin, normalled jacks, neon-vs-jewel, VHF detent, valve handwheel.
*** 2026-07-12 Sun @ 18:52:41 -0500 Added R54 vertical tape + R55 twin-needle gauge from the Ki-57 panel (gallery at 108)
Craig fed a Mitsubishi Ki-57 right-panel photo (instruments identified card by card, labels read from the Japanese: 速度計 airspeed, 遠方回転計 remote tach, 昇降計 VSI, 人工水平儀 artificial horizon, 高度計 altimeter, 気筒温度計 CHT, etc.). Two genuine gaps built: R54 vertical tape instrument (scrolling scale behind a fixed amber index — the Ki-57 remote tach's form, universal in glass cockpits; drag spins the tape) and R55 twin-needle gauge (mirrored FUEL/OIL half-scales, two needles one hub, per-half drag surfaces — the banked "twin mirrored gauge" from the French jet panel, distinct from N13 whose needles are read at their crossing). Cleanup pass earlier (c4fcee6): rotary-dial digits were painted under the finger wheel — now on top; legible demo defaults for pin matrix / voice loop / DSKY PROG; caption color unified; cam labels bumped. Reference filed (2026-07-12-ki57-right-panel.jpg). CDP-verified: tape drags (2400→1329), needles independent (fuel up leaves oil, oil down leaves fuel), zero exceptions.
*** 2026-07-12 Sun @ 18:59:08 -0500 Added R56 comfort-zone crossed needles (gallery at 109)
Craig fed a brass weather-station comfort meter and correctly picked it as different from both dual meters: N13 derives a NUMBER from the crossing (iso-curves), R55 reads two needles separately; this one lands the crossing in a printed categorical VERDICT (TOO WARM / TOO DRY / JUST RIGHT — active zone goes red, a digital advantage the paper face lacks). Brass knurled bezel, cream face, temp needle from bottom-left, humidity from bottom-right, per-half drag surfaces, needles clipped to the face (probe caught tails poking past the bezel at low values — fixed). Reference filed (2026-07-12-comfort-meter.png). CDP-verified: default 72F/45%/JUST RIGHT, warm and dry verdicts flip with the active label, zero exceptions.
*** 2026-07-12 Sun @ 19:13:41 -0500 Squash-merged to main (bc93388) + cleanup round 2
Squash merge of the 25-commit branch landed on main; branch deleted local+remote. Then a programmatic defect sweep (per-card overflow/empty/dead-readout/sparse bounding-box audit) caught R51's monitor bars streaking across the card — Chrome's CSS zoom miscomputes absolute left/right insets, fr tracks, AND stretched widths inside zoomed stages, so the keyset now uses fixed 52px grid tracks and fixed 40px flow bars (fixed px scale correctly under zoom; this is the third zoom-layout trap after the two in the same widget — noted for future stage-internal CSS: prefer fixed px inside .stagew). Seven intrinsically tiny widgets (toggle, chip, arm-to-fire, mini signal, ladder, thermometer, status lamp) got a .boost stage zoom (1.5x, compounds with the size toggle) so they stop floating in empty stages. Sweep false positives understood: clipped content (R43 horizon, R54 tape, N22 marquee, N25 counter) reports as overflow because getBoundingClientRect ignores clip; N13/N17/N28/R20/10 verified clean visually. Full regression green at 109 cards.
*** 2026-07-12 Sun @ 19:25:04 -0500 Page iteration: palette rebuilt as named instrument colors, moved to bottom; default size M
Per Craig: the palette left the top (now below Indicators) and stopped being a token dump — it is now a curated card of 36 NAMED colors grouped by role (Materials / Faces & inks / Lamps, LEDs & jewels / Screens & phosphors / Needles & controls), each with a name (Brass, Chart paper, Neon orange, Graticule green...) and where it appears on the instruments. No hex codes visible anywhere on the page (verified by innerText scan); page chrome (background, body text) deliberately excluded. Default widget size dropped 3x → M (1.7x). Data lives in a PALETTE array in gallery JS — the curation IS the data, since most instrument colors are local literals, not tokens.
*** 2026-07-12 Sun @ 13:01:27 -0500 Added the palette section
New "Palette — design tokens" section above Controls: 33 tiles read live from the generated =:root= rule via document.styleSheets (zero drift possible — what renders is what the widgets use). Hex tokens and glow rgb triples get swatches; mono/pulse-rate render as text tiles. Scales with the size toggle.
*** 2026-07-12 Sun @ 13:08:42 -0500 Added screen-color families + chips (first round)
Six period families defined (green = P1 phosphor built on =phos=, amber = P3 on the gold family, red = the nixie neon pulled out, blue = P4 white-blue, vfd = the marquee's =--vfd=, white = mono LCD), applied via scoped CSS vars with the shipped color as fallback — defaults are pixel-identical until a chip is clicked. Chips on the five clearly-screen widgets: R10 data matrix (amber), R17 CRT scope (green face), R19 waveform LCD (white), R31 radar (amber — the "in amber" one; green chip gives the classic PPI), N11 oscilloscope (green). Answers to the green question: the kit deliberately has four greens — phos #7fe0a0 (CRT trace), vfd #63e6c8 (marquee, blue-leaning), sevgrn #57d357 (LED), jewel-g #6fce33 (jewel lens); the phosphor-screen green is phos, not the others. Nixie left chipless: neon is only ever orange. Families live in gallery JS, not tokens.json — tokenize the winners once picked, the way amber earned its tokens. CDP-verified: recolors land on all five, interactions intact, zero exceptions.
*** 2026-07-12 Sun @ 13:15:19 -0500 Added spec sheets to all 84 cards
Every card now carries a collapsible "spec sheet" (a details element under the note) with up to 8 fields: input (always present, names the model — click ports everywhere including Emacs click-regions, drag needs a click/key idiom there, display widgets take no real input), solves, use (common vs specialty + where it shines), limits, origin, difficulty, prefer-when, and period (present on 41 of 84 — only where the component is clearly not timeless, e.g. nixie 1955-75 sits with keypads/telegraphs and clashes with VFD and LED seven-seg). Content lives in one INFO object keyed by card number; card() renders it. CDP-verified: 84/84 sheets present (missing-key check), fields render, zero exceptions.
*** 2026-07-14 Tue @ 02:00:49 -0500 Judged the two 2026-07-13 cross-needle references covered
The weather-station comfort meter is R56's form exactly (crossing lands in a printed categorical verdict; five zones vs R56's three is content, not mechanism). The SWR cross-needle meter is N13's defining mechanism (the crossing derives a number off printed iso-curves — forward/reflected watts → SWR). No new cards; both references stay filed in working/retro-stereo-widgets/references/.
*** DOING Widget validation pass
Craig walks all 109 cards; the lamps are his (click cycles off → amber → green, per-card localStorage key =gv-<no>=). Gate change with the option-1 approval (2026-07-12): the lamps no longer gate the widgets.js extraction (lossless, done) — they gate per-widget Emacs ports and the final catalogue blessing.

Runs as a *joint loop* (Craig, 2026-07-16): he walks a batch of 10-15 and reports card numbers + what's wrong; Claude fixes them in one pass, gates on the three probes, he reloads and re-walks. The lamps track progress only — they never recorded *what* was wrong with a card, so the defects live in this task body as they surface.

Statuses bake into the gallery source (=VSTATUS=, top of the lamp section) *periodically*, not once at completion as originally planned — localStorage is per-profile and dies with a cache clear, so at 8 cards baking is free and at 60 it's the difference between a record and a bad afternoon. Craig clicks "copy for source" under the index tally and pastes the block into the chat; the agent pastes it into =VSTATUS=. Precedence is localStorage → VSTATUS → off, so the live walk wins on the machine doing it and the baked record fills in on a fresh profile, after a clear, or on velox. Covered by =tests/gallery-probes/probe-vstatus.mjs=.

*The page can't copy for you; you press Ctrl+C* (settled 2026-07-16). Clicking "copy for source" drops the export into a selected textarea at the bottom-left of the page. Ctrl+C it and paste it into the chat. From a =file://= origin Chrome refuses both programmatic copy paths — =navigator.clipboard.writeText= rejects NotAllowedError, and =execCommand('copy')= returns *true while writing nothing*, which is worse because the page then claims a success it never had. A hand-typed Ctrl+C works fine, so nothing about the clipboard is broken: a Wayland-set clipboard crosses to X11 and Emacs reads it correctly (sentinel-verified against =gui-get-selection=).

*And the agent doesn't fetch it.* Never read the export with =wl-paste --primary=: PRIMARY holds whatever was last selected anywhere, so a blind read returns unrelated content (it surfaced a private SMS during this session). Craig pastes; the agent doesn't reach for it.

Progress: 8 green + 2 amber (R07, R52) of 109, baked 2026-07-16.
*** 2026-07-16 Thu @ 08:11:18 -0500 Ran the classification brainstorm — two empty cells and a missing axis
Brainstormed with Craig and wrote the result into [[file:docs/design/2026-07-12-control-grammars-reference.org][the control-grammars reference]] ("The brainstorm, run"). Crossed Foley's six elemental tasks against cardinality over the 109 cards.

Added *axis 8, set stability* (fixed at manufacture vs discovered at run time). The doc's original seven axes cannot find the kit's largest gap: axis 4 lumps all one-of-N together, so a chicken-head selector and a WiFi list share a cell that then reads as densely covered. The axis is invisible from inside hardware — every one-of-N in the kit has its positions engraved at manufacture — so a catalogue derived from period hardware inherits a blind spot exactly at dynamic sets, which is what its software consumers are made of.

Two empty cells, both wanted by live panels today: *select x one-of-N dynamic* (the bt device list, the net network list) and *text x alphanumeric* (the WiFi password; the kit enters digits three ways and text zero ways). Foley names six irreducible tasks; the kit serves five and a half.

Period models proposed rather than invented, per Craig's aesthetic gate: the answer to a changing set was never a control but a *re-labelable slot* — jukebox title-strip rack, Rolodex, scribble strip, switchboard strips. The slot is manufactured, the label is not.

Also found: the gallery cannot build its own chrome from its own kit (swatch chips, validation lamps, size toggle are all bespoke HTML; the size toggle duplicates card 06). That is a free completeness test worth re-running as the chrome grows.

Still open, recorded in the doc: the full seven-axis classification, the card-by-card audit behind the first-pass cell assignments, and the display side.

*** TODO Build the four widgets the taxonomy found :feature:design:
From the [[file:docs/design/2026-07-12-control-grammars-reference.org][taxonomy brainstorm]] (2026-07-16). Craig's gate: stick to the aesthetic — model a period control, don't invent a software-native one. Reference photo first per the usual pipeline, then judge, then card.

1. *Dynamic list* — model the jukebox wallbox title-strip rack (page-flip browse, rich slotted strips, select one). Fills the most damning cell; the bt and net panels both want it. Alternative model: Rolodex card spinner, better for long sets and weaker at showing several rows at once.
2. *Alphanumeric entry* — model the *industrial ABC-order keypad*. Craig's four references (=working/retro-stereo-widgets/references/2026-07-16-abc-keypad-*.png=, =-letter-drum-bank.png=) plus a survey superseded the teletype: the references are all ABC-order, not QWERTY, on stainless or membrane, with colour-coded function keys (yellow CANCEL, red CLEAR/NO, green ENTER/YES) that land on the kit's palette without translation. A faceplate, not furniture.
3. *Index typewriter* — stylus over a printed index plate, print lever commits (Hall 1881, AEG Mignon 1924). The survey's strongest find: select and commit on two separate controls is a grammar the kit has no exemplar of, and it is small and beautiful. Second card after the keypad.
4. *Chorded keyset* — six keys, no key per letter; the chord IS the character (Microwriter 1978, Perkins Brailler 1951). The most compact honest password field. The kit names the chorded grammar already (R50 is its safety form) but holds no chorded text control. Third, if a compact field is wanted.
5. *Swatch picker* — model the signal-lamp lens turret or theatrical gel wheel; a rotary whose detents are coloured lenses. Undecided whether it earns a card or stays page furniture.
6. *Legend switch* — the lit pushbutton whose cap legend and colour ARE the state, press to step (MIL-STD-1472 catalogs it; the doc already cites the standard). The validation lamps are one.

Items 1 and 2 fill the two empty cells and are wanted by live panels today; 3 and 4 are the survey's genuinely-new grammars; 5 and 6 come from the gallery-can't-build-itself finding and are lower stakes.

Banked from the survey, not cards yet: CDU/MCDU scratchpad + line-select keys (a staged-commit *flow*, not a new selector), joystick scroll-and-fire alphabet, multi-tap/T9 letter cycling (the most practical small-panel password entry), trackball gesture (Atari Quantum), keypunch program drum (IBM 029), Enigma lampboard (a *feedback* idiom — steal the 26-lamp grid as a readout, not as entry), Teletype Model 15 tape perforator.

Two flags carried from the survey: no true A-Z *thumbwheel switch* could be verified (Digitran/Grayhill mil-spec wheels are 0-9 or hex only), so the letter-drum reference is a *combination lock* and a card must say so rather than imply a switch that may not exist; and the trackball-gesture details are single-sourced.

*** TODO Taxonomy audit — verify the first-pass cell assignments :design:
The cross's cell assignments were read off card names and spec sheets, not audited card by card, so "confidence of completeness" isn't earned yet. Walk all 109 against elemental task x cardinality x set stability, cell the display side (left uncelled deliberately), and argue the four untouched axes (grammar, state authority, persistence, era). Would either confirm the two empty cells or find more.
** DOING [#B] Retro widget catalogue :feature:design:
:PROPERTIES:
:SPEC_ID: 3ac0d42c-db1a-4d21-bce4-e63785fef0ba
:LAST_REVIEWED: 2026-07-13
:END:
The panel widget gallery ([[file:docs/prototypes/panel-widget-gallery.html][docs/prototypes/panel-widget-gallery.html]]) grows into a retro-instrument component catalogue: reference photos of period hardware → gallery cards (the visual + behavioral spec) → reusable components for three targets (emacs svg.el, web/React, waybar). Tokens single-sourced in [[file:docs/prototypes/tokens.json][tokens.json]] (gen_tokens.py emits web/waybar/elisp); svg.el proof widget shipped (gallery-widget.el, needle gauge). Reference photos live in [[file:working/retro-stereo-widgets/][working/retro-stereo-widgets/]]. Collection converged at R56 (109 cards, all behaviorally verified; probes in [[file:tests/gallery-probes/][tests/gallery-probes/]]).

Build runs per the [[file:docs/specs/2026-07-12-component-generation-spec.org][component-generation spec]] (DOING; reviewed + decomposed 2026-07-12): web extraction first (ungated, lossless), then demand-gated Emacs/waybar ports. Banked variant/composition ledger lives in the 2026-07-11/12 session archive.
*** 2026-07-12 Sun @ 10:14:10 -0500 Wrote the component-generation spec (DRAFT)
[[file:docs/specs/2026-07-12-component-generation-spec.org][2026-07-12-component-generation-spec.org]] via spec-create: full spine (summary, problem, goals, two-altitude design, alternatives, 8 decisions with 1 open, 5 phases, acceptance criteria, readiness dimensions, risks). Gallery cited as the prototype evidence per the ui-prototyping rule (filed references + R01-R31 iteration history + final at 52a43ec). vNext items logged as the "Widget catalogue vNext" task.
*** 2026-07-12 Sun @ 20:24:37 -0500 Web library packaging approved — classic-script widgets.js + GW namespace
Craig approved with the componentization go-ahead (option 1): =widgets.js= as a classic script exposing one =GW= namespace, relative =<script src>= so =file://= keeps working, shared helpers inside, framework wrappers vNext. Decision flipped DONE in the spec (cookie 8/8). Gate per the same approval: extraction proceeds ungated (lossless transform); the validation lamps gate only per-widget Emacs ports and the final blessing.
*** 2026-07-12 Sun @ 20:57:50 -0500 Spec reviewed and decomposed (DRAFT → READY → DOING)
spec-review passed (Ready): 3 findings, all accepted and folded same pass — option-1 supersession of the demand-gated extraction order (Phases 1-2 swapped: extraction first, ungated), the card-record refactor named in Phase 1, stale counts refreshed (109 cards / R56). Probe baseline repaired first (753380e — two stale assertions from the evening sprint). Phase tasks below; =:SPEC_ID:= stamped on this parent.
*** 2026-07-12 Sun @ 22:56:40 -0500 Phase 1 complete — all 109 widgets extracted into widgets.js
All 109 card builders lifted into [[file:docs/prototypes/widgets.js][docs/prototypes/widgets.js]] (classic script, =GW= namespace, shared engine: svgEl/polar/vuDb/drag helpers/seg7/SCREEN_FAMS/gradient defs); every card is now a declarative record rendered by one =card()= path, all wiring blocks deleted. Widget CSS (incl. pulse/flipdrop/reelspin keyframes) moved from the gallery =<style>= block into =GW_CSS=, injected by widgets.js — verified pixel-identical under forced reduced motion (masked live-date cards N26/R35). Tick contract settled: the page owns the clock + demo signal; live meters expose value-driven handles; widget-owned animation lives in builders behind reduced-motion gates. Finale: slide toggle (card 01) is the first fully-realized component — its four option groups are =GW.slideToggle= constructor opts backed by =STYLES=, the gallery chips a demo rig on =handle.setStyle=. README consumers section documents the GW API + tick contract. Per-batch gate stayed green throughout (probe.mjs, probe-fams.mjs, a 239-check behavioral suite, reduced-motion smoke). Commits acee657 → 7b3bc47 (18 batches), all pushed.
*** TODO Phase 2 — demand inventory (Emacs/waybar) :design:
Widget-to-target matrix for the scripted-port targets: walk the live waybar panels (net/bt/audio/maint) and the Emacs surfaces Craig names; record which cards each actually wants. Lands in the spec's appendix; Craig approves. Tree untouched. Not =:solo:= — the matrix is his call.
*** TODO Phase 3 — Emacs ports of demanded widgets :feature:
Extend the =gallery-widget.el= pattern per demanded widget: ERT (tokens, geometry normal/boundary/error, SVG structure, state-tracks-value) + rsvg-convert side-by-side against the card; keymap/click-region interaction per widget; wired into =make test-elisp=. Gated on the Phase 2 matrix and Craig's green lamp per widget.
*** TODO Phase 4 — waybar pilot: audio panel :feature:dotfiles:
Restyle the audio panel's GTK CSS onto =tokens-waybar.css= + the banked composition idioms. Lives in =~/.dotfiles=; archsetup drives edit/test/commit/push end to end + inbox note. Visual result gets a manual-testing checklist entry (daily-driver panel).
*** TODO Phase 5 — Level-2 generator go/no-go :design:
After ~5 hand ports, weigh widget-level codegen with evidence (mechanical duplication vs judgment per port). Recorded as a dated decision in the spec; go spawns its own spec.
*** TODO Flip the spec to IMPLEMENTED
When the phases above close: status heading keyword → =IMPLEMENTED=, dated history line with the reason, Metadata =Status= mirror. Three lines, one file.
** TODO [#C] Maint live-refresh hairline replacement :feature:maint:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-14
:END:
From the roam inbox (routed 2026-07-13): the memory-killer section seemed to update too often, and "it's a bit unclear what the line is doing; consider something else." Diagnosis (2026-07-14): the data cadence is already the requested 3s (gui live tier, _LIVE_SECONDS); the perceived churn is the live-refresh hairline — the 2px bar under the live sections that drains full-to-empty over each 3s window, redrawn at 150ms (gui._hair_tick, viewmodel.refresh_fraction). It exists to tell a stale board from a frozen one (2026-07-09), but it reads as constant unexplained motion. Design call for Craig: replace the draining line with something whose meaning is legible — candidates: a dot that blinks once per refresh, a "3s" age caption that only appears when refresh is overdue, slowing the drain redraw, or dropping the indicator on live tiers and keeping it only when data goes stale. Keep the stale-vs-frozen distinguishability that motivated the hairline.

** TODO [#B] Net doctor expansion v1 — VM live verification :feature:dotfiles:network:
:PROPERTIES:
:SPEC_ID: ce29b103-ed9d-4f56-bf8c-9ed8fe680ff3
:LAST_REVIEWED: 2026-07-13
:END:
Build the [[file:docs/specs/2026-07-11-net-doctor-expansion-spec.org][net doctor expansion]] (IMPLEMENTED). Adds the control-plane cluster (rival-manager / nm-masked / keyfile-perms) and a sharper auth verdict to the shipped net doctor (=~/.dotfiles/net/=). Archsetup owns the dotfiles work end to end — edit, test, commit, and push in =~/.dotfiles=, then drop an inbox note. All build phases shipped and fake-verified; the one open piece is the VM live verification below.
*** 2026-07-11 Sat @ 02:47:47 -0500 Built the read-only control-plane probe
New module =net/src/net/control_plane.py= plus =diag.py= wiring, on dotfiles main (=21ca3ff=, pushed). =net diagnose= now carries a =control_plane= key in its envelope with three read-only signals: NetworkManager state (masked/failed/stopped/active, finer than the plain =is-active= the ladder used), any rival manager active alongside NM (dhcpcd, systemd-networkd, iwd via =systemctl is-active=), and the active profile's keyfile permissions (=stat= under an env-overridable =NET_NM_CONNECTIONS= root, default =/etc/NetworkManager/system-connections=). Detection only: no classifier change, no new step, no bearing on =overall=. Surfaced via =net diag --json=. The masked/failed read also runs in the nmcli-down early-return path, or a masked NM would short-circuit before the probe. Every read bounded through =cmd.run='s timeout; degrades to unknown / no-rival / None on an unreadable tool or unstattable file, so a probe that can't see never invents a fault. As a normal user the real 0700 root:root dir is unstattable, so the keyfile signal reads unknown (readable only under root, e.g. the doctor's privileged path). 21 new tests, full =make test= green; =/review-code= approved (two Minor items, both Phase 1 concerns already tracked in the spec: link-scoping the rivals and guarding the keyfile path in the chmod fix). Inbox note sent to dotfiles.
*** DOING [#B] Phase 1 — control-plane verdicts + privilege model
Unblocked now that panelkit ships. =classify= gains =rival-manager=/=nm-masked=/=keyfile-perms= (all =fixable=), ordered ahead of the generic not-running rule, with the masked check reachable in the NM-down early-return path. Fixes register as =VERBS= + =ACTIONS= (=disable-rival=/=unmask-nm=/=chmod-keyfile=), each narrowly scoped. Live verification needs a real rival/masked/bad-keyfile state (a VM or a deliberately-broken host), so not solo.
**** 2026-07-11 Sat @ 04:32:06 -0500 nm-masked verdict shipped + first panelkit integration
On dotfiles main (=d73eba6=, pushed). The clean, unambiguous verdict of the three, done first to prove the panelkit-in-net rails: =classify= reads =control_plane.nm_state=; on =masked= it returns a fixable =unmask-nm= verdict (=remedy_class="privileged"=) ahead of the generic nm-service rule (=nm-restart= can't start a masked unit). New priv verb =unmask-nm= (=systemctl unmask NetworkManager=), =repair_unmask_nm= (unmask → start → verify), narration entry. The doctor resolves a privileged verdict through =panelkit.resolve()= before running: can't-elevate (GUI, no passwordless, no tty) degrades to the manual guide instead of attempting; existing tiers (no remedy_class) run unchanged. The =net= shim + test add =panelkit/src= to sys.path. Design call shipped (flagged for Craig): CLI =net doctor --fix= is the deliberate act satisfying the Confirm floor, so no CLI prompt added; panelkit contributes the run/prompt/guide degradation on the CLI; the GUI arm-press stays its confirm (panel wiring pending). 765 net tests + 65 suites green, review-code clean. Inbox note sent.
**** 2026-07-11 Sat @ 05:16:24 -0500 rival-manager + keyfile-perms verdicts shipped
On dotfiles main (=c9f8604=, pushed). Craig approved proceeding with the default precedence. =classify= adds =disable-rival= (rival dhcpcd/networkd/iwd active + a failing link) and =chmod-keyfile= (bad-perms keyfile + a failing link), both =remedy_class="privileged"= through the same panelkit gate. PRECEDENCE settled: both gate on a local-link symptom (never nag next to a working link), sit behind rfkill/service (more fundamental) and ahead of the generic reset (naming the cause beats a blind reconnect); tested (rfkill beats rival, rival beats reset). Priv verbs =disable-rival= (three-unit allowlist), =chmod-keyfile=/=chown-keyfile= (paths validated under system-connections, no traversal). Repairs re-derive their target (=active_rivals= / the active connection). =fake-systemctl= gained a state dir so a disable flips the re-check. 782 net tests + 65 suites green, review-code clean. Two coverage edges deferred to the VM: the keyfile fix's root-owned pass (only reproduces under root), and whether disable-rival needs a follow-on reset to restore the link. All three Phase 1 verdicts now fake-complete; VM live-verification is the remaining piece (see the sub-task).
**** TODO Net Phase 1 live verification — run the scenario harness against a VM
The harness is BUILT (archsetup =a364c1e=): =scripts/testing/net-scenarios/{10-nm-masked,20-rival-manager,30-keyfile-perms}.sh= (break/fix/assert + the expected diagnose verdict) and =run-net-scenarios.sh= (rsyncs net+panelkit into a target over ssh, drives the scenarios). Scenario logic reviewed; the ssh transport plumbing is UNEXERCISED (marked first-draft) and wants a shakeout on the first real run. What remains needs Craig: a booted VM (a real kernel + network stack — NM does NOT run reliably in nspawn, so a container only shows the mask-symlink/keyfile-mode half, not "NM active") with NetworkManager + dhcpcd installed, reachable over ssh as root. Then =run-net-scenarios.sh --target root@HOST --profile <saved-profile>=. The two live questions: disable-rival was already chained to a follow-on reset (=be15a81=), so the run confirms whether the reset fires or NM auto-recovers; and the keyfile root-owned pass (only reproduces under root). The by-hand equivalent is the "Manual testing and validation" → "Net doctor privileged fixes" checklist.
*** 2026-07-11 Sat @ 06:31:29 -0500 Built the sharpened auth verdict
On dotfiles main (=12e3e76=, pushed). =gather_context= derives an auth cause on an auth failure from the saved profile's key-mgmt + the scanned SECURITY flags for the SSID (not GENERAL.REASON, which only marks *that* auth failed): sae / hidden / enterprise / generic. =classify= turns sae and hidden into =fixable= Privileged profile-modifies (=key-mgmt sae= + PMF, or the hidden flag) routed through =panelkit.resolve()= like the Phase 1 control-plane verdicts; enterprise and generic stay =needs-user-action= with a cause-named message. Two new priv verbs (=conn-sae=, =conn-hidden=, uuid-validated single =nmcli connection modify= commands) + =repair_auth_sae=/=repair_auth_hidden=, each chained into a reset (FIX_CHAINS) so NM reconnects with the corrected profile. Hidden fires only on a non-empty scan missing the SSID, so a failed scan can't fabricate it. Tightened the =fake-nmcli= key-mgmt hook to require the =-g= prefix so the conn-sae modify isn't read back as a get_value. Pairwise (reason × profile-state) covered in =TestAuthCauseDerivation=. 813 net tests + 65 suites green, review-code Approve, voice. Inbox note sent. Live half (real WPA3/hidden profile-modify) is the VM/manual checklist.
*** 2026-07-12 Sun @ 09:14:00 -0500 Flipped the net spec to IMPLEMENTED and logged the vNext items
Spec status heading now IMPLEMENTED (dated history line + Status mirror); all four phase headings DONE. vNext items (flaky/drops cluster, DoT/DNSSEC verdict, profile hygiene) logged as the "Net doctor vNext" task. The privileged-fix live halves remain with the VM live-verification sub-task and the manual-testing checklist — findings there come back as bugs.

** TODO [#C] Net panel speedtest history :feature:dotfiles:network:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-14
:END:
From the roam inbox (routed 2026-07-13): the networking panel should track speedtests over time with appropriate info. Shape: persist each SPEED TEST result (timestamp, down/up, latency, server) to a small local store and surface history in the net panel. Design questions for work time: retention window, which fields matter, and presentation within the panel's ~400px width (recent-results list vs trend readout). Point-in-time results exist today; the gap is comparison across days and venues.

** TODO [#D] Net doctor vNext :feature:dotfiles:network:
Deferred from the [[file:docs/specs/2026-07-11-net-doctor-expansion-spec.org][net doctor expansion spec]] (IMPLEMENTED, v1 shipped): event-log correlation for the flaky/drops cluster (powersave, roaming stalls, USB autosuspend, no-reconnect-after-resume, firmware crashloop drop signatures — needs history a one-shot probe can't see); a DoT/DNSSEC-specific verdict distinguishing "the venue resolver mangles DNSSEC" from the generic DNS-not-resolving; per-profile autoconnect/duplicate-profile hygiene.

** TODO [#D] Bt doctor vNext :feature:dotfiles:bluetooth:
Deferred from the [[file:docs/specs/2026-07-11-bt-doctor-expansion-spec.org][bt doctor expansion spec]] (IMPLEMENTED, v1 shipped): the stale-bond re-pair-offer signature (v1 keeps re-pair strictly user-initiated; the signature must be designed and validated before the doctor ever offers it); a connection-parameter/coexistence hint tail for a "keeps dropping" verdict; the bt-audio-profile expansion beyond the current a2dp repair (codec fallback, default-sink, absolute-volume) — wants coordination with the audio doctor so the two panels don't claim the same A2DP/HFP diagnosis with divergent verdicts.

** TODO [#D] Widget catalogue vNext :feature:design:
Deferred from the [[file:docs/specs/2026-07-12-component-generation-spec.org][component-generation spec]] (DRAFT): framework wrappers for the web library (React or otherwise — demand-gated, none until a real framework consumer exists); Level-2/3 widget codegen if the spec's Phase 5 decision point says go (a go spawns its own spec); ports beyond the demand matrix as new consumers appear.

** TODO [#D] Maintenance console vNext :feature:
Deferred from the [[file:docs/specs/2026-07-07-maintenance-console-spec.org][spec]] (v1 ships determinate remedies only): AI/workflow assistance on read-only metrics (the vLater tier — failed-unit diagnosis, journal-error fix suggestions, OOM investigation); SIGKILL escalation for TERM-survivors on the KILL lever; live streaming meters on evidence rows where a count could be a level.

** TODO [#D] Retention-repair remedy unreachable from the panel GUI :bug:dotfiles:
snapshot_retention_repair (WRITE SANE LIMITS — =snapper -c <config> set-config TIMELINE_CREATE=yes TIMELINE_LIMIT_*=<TOML values>=) sits in the maint remedy table but is wired into no GUI layer (no reference in panel.py, viewmodel.py, or gui.py) — it's reachable only via =maint doctor review= / =maint fix=. Worse, the REVIEW & FIX roster's wording for item-bearing remedies says "per-item — on its subpanel", which for this remedy points at a key that doesn't exist. Found 2026-07-08 while walking the SNAPSHOTS subpanel with Craig.

Two fix shapes, decide at work time: (1) wire a digest key — e.g. on the timeline row when the config's installed TIMELINE_LIMIT_* drift from the TOML values (needs the probe to read the installed limits for comparison); or (2) keep it CLI-only deliberately and special-case the roster wording for remedies with no panel key ("CLI only — maint fix"). Option 2 is a few lines; option 1 makes limit drift visible on the board, which is the 2026-05-26 pile-up's root cause. Severity minor × rare edge = P4 per the bug matrix.

** TODO [#C] zfs base VM image build failure: ZFS DKMS module missing :bug:zfs:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
=FS_PROFILE=zfs make test-vm-base= fails inside the VM at initramfs time: archangel reports "ZFS module not found! DKMS build may have failed" against the installed kernel (linux-lts 6.18.38 at the 2026-07-08 attempt). Consequences: the maint scenario harness's zfs lane (Phase 12) is filtered but unexercised, and a real zfs bare-metal install via archangel would plausibly hit the same wall. Priority per the bug matrix: Major severity (zfs install path broken) × some-users-sometimes = P3. When fixed, run =FS_PROFILE=zfs bash scripts/testing/run-maint-scenarios.sh --list= and add zfs scenario files (zpool scrub / autotrim / snapshot destroy) to the harness.
*** 2026-07-14 Tue @ 01:40:48 -0500 Diagnosed: OpenZFS/kernel version skew, blocked on archzfs
Reproduced in ~1 minute of install: =dkms install zfs/2.3.3 -k 6.18.38-2-lts= exits 1 during pacstrap. Root cause confirmed: OpenZFS 2.3.3's META declares Linux-Maximum 6.15, and the VM installs linux-lts 6.18.38. The first release supporting 6.18 is 2.4.0 (2.4.1 covers 6.19), and archzfs currently serves only zfs-dkms 2.3.3-1 — nothing on our side to fix. Unblock condition: archzfs publishes zfs-dkms ≥2.4.0; recheck with =curl -s https://archzfs.com/archzfs/x86_64/ | grep -o 'zfs-dkms-[0-9.]*'=, then rerun =FS_PROFILE=zfs make test-vm-base=.

** TODO [#B] Consistent keybinding family for the panel console :feature:hyprland:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Consider putting every panel (net, bluetooth, audio, timer, and the coming maintenance console) on one consistent chord family — a shared modifier set (Super+Shift, Control+Alt, or similar) plus a mnemonic letter per panel (N/B/A/T/M). Today the panels open via waybar clicks only; a uniform chord family makes them keyboard-reachable and predictable. Watch for collisions with existing binds: Super+Shift+A is already PTT toggle, and the hold-to-talk grave bind is load-bearing. Decide the family, audit current hyprland binds for conflicts, wire via the dotfiles hyprland config, and document in the keybind reference. Both machines (velox can't QMK-remap, so chords must work on a plain laptop keyboard).
*** 2026-07-14 Tue @ 00:31:36 -0500 Folded Craig's ask for a maintenance-panel keybinding; bumped [#C] → [#B]
Craig asked (in session, 2026-07-14) for a maintenance keybinding specifically — the panel he's reaching for without one. Maintenance (M) is the priority chord when this task gets worked. The capture graduated the task from parking lot to active backlog.

** DOING [#B] Run-time privilege model, standard across every panel doctor :feature:dotfiles:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-13
:END:
The audio input/output doctor is gaining a run-time privilege model (see [[file:docs/specs/2026-07-10-audio-doctor-input-side-spec.org][docs/specs/2026-07-10-audio-doctor-input-side-spec.org]], decision "The doctor may use sudo, resolved by context at run time"). Craig's call, 2026-07-10: make it a standard, "revise the other panels to be consistent with these changes."

The model: a doctor resolves its privilege at run time from three signals — passwordless sudo available (=sudo -n true=, which never hangs), a tty to prompt at, and whether it is the GUI panel. Four remedy classes: Auto (user-scope, reversible), Privileged (needs sudo — runs where passwordless, prompts on a CLI tty, degrades to Guide in a GUI with neither), Reboot-tail (run the applicable part, then instruct the reboot), and Guide (physical/BIOS/wait-for-upstream, nothing to run). Safety floor: every Privileged and Reboot-tail remedy defaults to Confirm or Arm tier, never silent Auto, because passwordless sudo is not consequence-free.

The shared helper is built (see the dated entry below). What remains is per-panel adoption: wire each doctor's remedies through =panelkit.privmodel.resolve()= and audit them against the Confirm/Arm floor, and reconcile maint's =priv.py= build/fire table with the model rather than leaving its implicit always-passwordless assumption. That wiring lives in the per-panel fix phases (net Phase 1, bt Phase 2, audio input/output), each needing a real privileged host to verify =--fix= end to end, so none is agent-solo.

Craig's decision, 2026-07-12: maint's harmless-reclaim privileged remedies (the silent CLEAN UP set — paccache keep3, journal vacuum, coredump clean) STAY silent-auto. The reconciliation gives that class a sanctioned, documented exception to the confirm floor rather than forcing Confirm/Arm; the value of the floor holds for everything else. Where sudo is not passwordless, maint should degrade per the model (prompt on a tty, guide in a GUI) instead of hard-failing.

*** 2026-07-11 Sat @ 03:48:57 -0500 Built the shared privilege model (panelkit)
On dotfiles main (=b987820=, pushed). Craig chose to share a library, not per-panel copies. New =panelkit= package (=panelkit/src/panelkit/=): pure library code in the dotfiles tree, no CLI, no stow package — a consumer reaches it by adding =panelkit/src= to =sys.path= alongside its own =<panel>/src= and =from panelkit import privmodel=. Self-contained (its own =cmd.py=) so a panel plus panelkit stays a coherent unit (keeps the pluggable-panel option open). =privmodel=: four classes (=auto=/=privileged=/=reboot-tail=/=guide=), =resolve(remedy_class, ctx)= → a =Resolution= (=RUN=/=PROMPT=/=GUIDE_ONLY=, =confirm=, =reboot_after=, =reason=) over a =PrivContext= (passwordless via =sudo -n true=, tty, is-GUI). Safety floor enforced and property-tested across all 8 context combos: every privileged/reboot-tail remedy that runs requires a deliberate confirmation, never silent auto. =PANELKIT_SUDO= test seam (="true"=/="false"=), =detect_context()= + =plan()= convenience. 19 tests, full =make test= green (65 suites), review-code clean. Inbox note sent to dotfiles. NOT wired to a consumer — that's the per-panel adoption above.

*** 2026-07-12 Sun @ 09:01:25 -0500 Reconciled maint onto the shared privilege model (dotfiles 7937a9e)
maint's doctor now resolves every privileged step through =panelkit.privmodel= instead of the old always-=sudo -n= hard-fail: RUN where passwordless (unchanged), PROMPT on a CLI tty (plain sudo, preceded by a wall note), GUIDE events in a GUI with neither — the typable command on the wall, nothing executed. User-scope steps never resolve; a macro's user steps still run around guided privileged ones. Craig's silent-auto decision (above) is implemented as the documented exception: recorded in =maint/remedies.py= (tier semantics) and =panelkit/privmodel.py= (the floor's own docstring), waiving only the confirm gesture, only for the auto-tier reclaim set. TDD (15 new tests red→green incl. both wall renderers), full suite 65/65, live-verified RUN + GUIDE end-to-end on this host via the real launcher shim. Bonus root-cause fix caught by the live run: =probes/logs.py= crashed on journalctl MESSAGE=null entries (key present, so the =.get= default never applied), which broke every real fix's re-probe — normalized at =_journal_lines= with a regression test. Remaining under this task: net Phase 1 / bt Phase 2 / audio adoption (queue items 2-3), and the PROMPT-path + GUI-wall manual checks (see Manual testing and validation).

** TODO [#B] Scrolling/Carousel layout: frame fit + wrap-around :hyprland:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-06
:END:
Disabled 2026-06-12 (bind and cycle entry points removed; Super+Shift+S reassigned to whole-desktop screenshot). The layout needs real work before it earns its chord back:
- What fits in each frame: column/frame sizing so windows land at usable widths instead of arbitrary slices.
- Wrap-around: navigating past the last frame should wrap to the first (and vice versa).
- Whatever else surfaces in daily use once the above land.

The support machinery was deliberately kept for this task: =layout-navigate= and =layout-resize= retain their scrolling branches, =waybar-layout= still renders the scrolling state, and the unbound legacy =cycle-layout= script still lists it. Re-enabling is two lines: add =scrolling= back to =LAYOUTS= in =layout-cycle= and restore a direct-jump bind (the old chord is taken now — pick a new one). The =tests/layout-cycle= suite pins the disabled state and will go red on re-enable, which is the reminder to update it.

** TODO [#C] Dotfiles stow conflicts: first-launch risk + restow directory handling :bug:dotfiles:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-14
:END:
*** 2026-07-14 Tue @ 00:51:51 -0500 Ratio calibre check passed; waypaper canonical decided (dark-lion)
Ratio's ~/.config/calibre is a directory symlink into the dotfiles repo (stow folded the whole dir), so the first-launch gap never existed there — check closed. Craig decided dark-lion.jpg is the canonical waypaper wallpaper; the repo config.ini updated from the that-one-up-there.jpg placeholder (the file is skip-worktree volatile, unskipped for the commit and re-flagged). Remaining: when velox is back online, run make conflicts / make reset there so its old conflict copy clears against the updated repo.
*** 2026-07-02 Thu @ 17:30:00 -0400 Shipped the Makefile hardening + first-launch guard (dotfiles 42a82d2)
The solo-able subset landed in the speedrun. =make conflicts <de>= is the loud first-launch guard: dry-runs all tiers, parses all four stow error shapes (plain file conflict, foreign symlink, dir-over-file, and restow's unstow_contents non-directory ERROR), lists each blocker with a directory/foreign-symlink marker, exits 1 when any exist. =make reset= now pre-clears the directory and foreign-symlink blockers =--adopt= aborts atomically on (removals printed; repo version wins per the target's contract), then adopts + git-checkouts as before. =make restow='s overwrite path switched rm -f → rm -rf so directory conflicts clear. 8 sandbox tests drive the real Makefile against a throwaway HOME (44 suites green). Also verified on velox: the whereami and mpd-playlists conflicts noted in this task were already hand-converted 2026-06-29 — =make conflicts hyprland= reports clean live. REMAINING (deferred per Craig's speedrun pre-flight): the waypaper canonical decision (live velox dark-lion.jpg vs repo that-one-up-there.jpg) and the ratio calibre-symlink check (ratio paused).
From the velox calibre incident (2026-06-27, note in ~/.dotfiles/inbox/processed/): calibre was launched before =make stow= ran, wrote its own default config into =~/.config/calibre/=, and silently blocked its own stow — it ran on factory defaults while the rest of common/ stowed fine. General pattern: any GUI app that auto-creates config on first run, launched before stow, blocks its own stow the same way. Velox was repaired by hand (=ln -srf= symlinks byte-identical to =stow --no-folding= output).

Remaining work (re-graded C 2026-07-02 — the first-launch risk and the Makefile handling shipped in the speedrun; what's left is a decision and a paused-machine check):
- Waypaper canonical decision (Craig): live velox has =dark-lion.jpg=, repo has =that-one-up-there.jpg= (placeholder?). Decide the canonical, then =make reset= clears it.
- Ratio check: whether =~/.config/calibre/*.json= on ratio are symlinks or real files — same first-launch gap likely if calibre ever launched there before stow. On the ratio trip list.

** TODO [#B] Audit dotfiles/common directory :chore:dotfiles:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-14
:END:
Refiled from the archsetup task audit (2026-06-28), landed via ~/.dotfiles/inbox; the dotfiles content split into its own repo 2026-06-16 but the task tracking stays here per Craig (2026-07-02). Three parts:
- Review all 50+ scripts in =~/.local/bin= and remove unused ones.
- Check dotfiles for uninstalled packages and remove orphaned configs.
- Verify all stowed files are actually used.
*** 2026-07-14 Tue @ 01:40:48 -0500 Built the audit evidence report
Shipped as =docs/2026-07-14-bin-audit-evidence.org= in the dotfiles repo (260752a). 150 scripts bucketed: 69 keep (referenced or cron-driven), 74 kill candidates (zero references in the tree), 7 flagged (all dwm-tier, expected on a hyprland host). Config sweep: audacious and wofi configs are orphan candidates, ranger needs an install-vs-delete call (declared in archsetup but not installed on ratio). Shell history was too shallow (~700 lines) to prove by-hand disuse either way — the kill pass stays Craig's call in the parent task.

** TODO [#B] Waybar network module — custom/net :feature:waybar:network:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Unifies the old wifi-no-internet indicator (was =[#C]=) and the network-manager
dropdown (was =[#B]=) into one =custom/net= module: a tested Python =net= engine
(nmcli + diagnostics), a thin bar indicator, and a GTK4 layer-shell panel. Code
lives in the dotfiles repo (hyprland tier + a =net/= package like pocketbook);
archsetup only installs deps. Secrets stay in NetworkManager's own store (no
separate credential store). The =captive= script becomes the diagnostics engine.
Full design, acceptance criteria, and the failure-mode coverage table:
[[file:docs/design/2026-06-29-waybar-network-module-spec.org][2026-06-29-waybar-network-module-spec.org]].

Phases below, dependency order. Engine/unit work is agent-verifiable (=unittest=
+ fakes on PATH, coverage via venv); the live-network and visual states need real
conditions, filed under "Manual testing and validation".

*** 2026-06-29 Mon @ 20:19:11 -0400 Phase 1 shipped — indicator + console recovery
Shipped to the dotfiles repo (10 commits, =5254bd8=..=c095a22=, pushed to main).
The =net= engine is a src-layout Python package in-tree, imported by a bin shim
that resolves the stow symlink back to the repo — so it runs from a bare TTY with
no install, which the recovery path depends on.

Landed: =net status= (fast path, one nmcli call + sysfs, degraded fallback in
budget) + =net probe= (native captive probe, single-flight flock, atomic cache,
fresh/stale/expired/unknown classes, iface/SSID/UUID invalidation); =waybar-net=
replacing =custom/netspeed=, throughput → tooltip, CSS states in both themes +
live; =net diagnose= (read-only steps) + =net repair= (rfkill/reset/bounce/
dns-test, cleanup-verified) + =net doctor [--fix]= with the four terminal
classifications; =net portal= + the =captive --probe-json= refactor; redacted
JSONL event log; Makefile recovery targets (=make online= etc.); =~/.config/net/
config=. Verified live: =make net-status= reads the real wlp170s0 / @Hyatt_WiFi.

Airplane (Craig's call, option 1): =custom/net= absorbs only the *display* — net
reads the airplane-mode state file and shows an airplane state/glyph. The
airplane-mode toggle stays (it's a low-power mode — radios + CPU + brightness +
services — not a radio switch), now on =custom/net='s right-click + signal 15.
Deleted: =waybar-airplane=, =waybar-netspeed=, =custom/airplane=, their tests +
css. =airplane-mode= kept.

Tests: 160 in =tests/net/= (fake nmcli/curl/rfkill/resolvectl/ping/getent/
systemctl on a temp PATH; doctor-classification fixtures; degraded-under-slow-
nmcli benchmark) + the =captive= probe-mode tests; full dotfiles suite green (32
suites). Coverage-gap pass via throwaway venv: pure modules ≥90% branch
(classify 100%), IO-error branches excused in the test docstring.
Deferred to Phase 2/3: archsetup deps (gtk4-layer-shell/python-gobject Phase 2,
speedtest-go-bin Phase 3 — not added before the code that needs them).
Verify (manual, live): see Manual testing and validation.

*** 2026-06-29 Mon @ 22:19:25 -0400 Phase 2 shipped — panel shell + connection management
Shipped to dotfiles (commits =4e7740f=..=24bcac5=, pushed). Engine: =net list= (saved
MRU + in-range wifi scan, infrastructure types filtered), =net up/down= (UUID-keyed,
mutation safety — keep prior link until target activates, classify wrong-password vs
generic, report auto-reactivation), =net add/edit/remove/rescan= (open + WPA-PSK;
enterprise activate-only; secret to NM's store, never our JSON/log — tested).

Panel: a GTK-free PanelModel (selection, four state machines, the UX-flow enable
rules, terminal states) + a GTK4 gtk4-layer-shell window (=net panel=) anchored
top-right under the bar — Connections section with MRU list, active marked, signal
glyph, row-click select, Connect/Add/Forget/Rescan, confirm-on-forget, worker-thread
engine calls via GLib.idle_add. GTK imported lazily so the CLI/tests stay GTK-free.

Bar interactions (settled with Craig over live iteration): left = =net-panel= toggle,
middle = =net portal=, right = =net-fix= (notify the doctor result when one-way; open
a terminal only when the outcome is fixable — the sudo/interactive case). Airplane on
Super+Shift+A. archsetup adds =gtk4-layer-shell= + =python-gobject= (this commit);
already on velox.

Tests: 204 in tests/net (merge ordering/dedup, up/down mutation safety, no-secret-leak
on add/edit, panel model + state machines, gui row-format helpers). Full dotfiles suite
green (32 suites). Live-verified on velox: panel opens/toggles, list shows real 24
profiles, right-click notification delivers (Craig confirmed). Phase 3 (diagnose/repair/
speedtest IN the panel) is next; the engine for it already exists from Phase 1.

*** 2026-06-29 Mon @ 22:43:40 -0400 Phase 3 shipped — diagnostics + speed test in the panel
Shipped to dotfiles (=91277cf=..=691abcb=) + archsetup (=48052d6=, speedtest-go-bin),
pushed. Engine: =net speedtest= (parses speedtest-go --json → ping from latency ns,
down/up from per-server byte rates; missing-backend / offline / malformed → error
envelope per the failure table). Panel grew a section switcher with four pages:
- Connections (Phase 2).
- Diagnose: =net diagnose= on a worker thread, each step a row (✓/✗/… glyph + title +
  redacted evidence), read-only; Open-portal button when captive.
- Repair: "Get me online" (=net doctor --fix=) + tiers (rfkill/reset/bounce/dns-test)
  + force portal. Confirmations in-panel with the spec's exact wording; the privileged
  tiers run via =net-popup= terminal (where the sudo prompt + step output, incl.
  cleanup-verified, show) — a panel has no tty, and pkexec would mean a prompt per op.
- Speed test: in-process =net speedtest= (no privilege → inline result: ↓/↑ Mbps + ping
  + server), Run/Cancel (Cancel pkills the child), error envelope shown.

213 net tests; pure helpers (step_indicator, format_speedtest) unit-tested. Full
dotfiles suite green (32 suites). One unverified assumption: speedtest-go's dl/ul unit
(taken as bytes/s; =BYTES_PER_SEC= flips it) — needs one real run vs a reference. The
in-panel repair streaming (vs terminal) is a named future polish once the GUI-privilege
story settles.

The waybar network module ([#B] parent) is now COMPLETE through Phase 3. Phase 4
(in-app help + user guide) and Phase 5 (VPN/WireGuard) remain as future work; the core
feature (indicator + recovery + panel + diagnostics + speed test) is done.
Verify (manual, live): see Manual testing and validation.

*** 2026-07-09 Thu @ 16:32:54 -0500 Audit reconcile: Phase 4 is filed on the dotfiles side, waiting on them
The dotfiles project accepted the Phase 4 handoff and filed it as a =[#C]= task in their own =todo.org= (their note, 2026-07-08 16:56): the help-text audit + panel help affordance, the user-guide/README, and the ratio rollout doc. Not started there. They ping when it lands, and this task's Phase 4 child closes then. Nothing to do here meanwhile.

*** TODO Phase 4 — docs + rollout :network:blocked:
Deliverable: in-app help (=net --help= + per-command, panel help affordance);
README/user-guide (commands, indicator states, panel, config keys, make targets,
troubleshooting from the failure table, rollback); archsetup Hyprland dep install
(=gtk4-layer-shell=, =python-gobject=, =speedtest-go-bin=); ratio manual dep +
stow step.
Verify: =net --help= and each subcommand complete; user-guide covers every command
+ the recovery targets.
Build handed off to the dotfiles project 2026-07-04 (=~/.dotfiles/inbox/2026-07-04-1305-from-archsetup-phase4-handoff.md=): archsetup deps confirmed installed, the remaining help/user-guide/rollout-doc work is in the net package. dotfiles pings back when it lands.

*** TODO Phase 5 — VPN / WireGuard CLI fold (vNext) :network:
Rescoped 2026-07-04 (audit): the tunnels track already shipped most of the original Phase 5. Panel tunnel bring-up/down and detection landed (dotfiles 2d9d060 probes tailscale/NM-wireguard/Proton; 21db05a brings overlays up/down from the panel's Tunnels sub-view; 31ba056 diagnose/doctor understand tunnel routes; archsetup 2e40781 wireguard config import; the net-panel-other-interfaces spec is IMPLEMENTED). What remains for Phase 5 is only the =net vpn ...= CLI subcommand — cli.py still has no vpn/tunnel parser. Fold the panel's existing tunnel operations into a CLI surface; spec separately when picked up.

** TODO [#C] Waybar collapse control: replace the triangle glyph :feature:waybar:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-14
:END:
From the 2026-07-04 roam capture. The waybar collapse mechanism (click the triangle, the bar sections redisplay shortened) works, but the triangle glyph doesn't match the instrument-console aesthetic the panels now use. Replace it with something in keeping with the console look. Aesthetic decision — bring Craig two or three concrete glyph/style options (a machined chevron, a console-key style expander, an engraved caret) before wiring. Dotfiles waybar config (handled per the archsetup-owns-dotfiles rule). Raised alongside the net-panel/audio speedrun; deferred from it because the glyph choice is a taste call.

** TODO [#C] Net panel: driver-health diagnostic tier :feature:network:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-14
:END:
Follow-up from the 2026-07-04 net-panel hardening speedrun (Craig's cj question on the no-WiFi item). The shipped no-wifi-hardware verdict covers "no adapter at all." This tier covers "adapter present but the driver is wedged": read-only health signals — =ip link= (device present but no-carrier / down), =dmesg= / =journalctl -k= for firmware-load failures, =rfkill= for a hard block, =modinfo= / =lsmod= for the driver module — classified before a generic reset. Remedy actions: a privileged =modprobe -r <mod> && modprobe <mod>= reload of the wifi driver, and a firmware-package pointer when the failure is a missing/failed firmware load. Dotfiles net-package work (handled per the archsetup-owns-dotfiles rule). Design pass first to decide whether it's worth a repair tier vs a needs-user-action pointer.

** TODO [#B] Desktop-settings dropdown panel :feature:waybar:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Initial spec written 2026-07-02: [[file:docs/specs/2026-07-02-desktop-settings-panel-spec.org]] (DRAFT — four decisions await Craig's review before build; architecture updated to the net panel's Blueprint/GTK4 stack).

One waybar dropdown gathering the desktop toggles and sliders into a single settings panel, opened from a gear/settings glyph on the bar. Incorporate:
- *Auto-dim* toggle (the =custom/dim= feature just shipped — fold in here, or keep the standalone indicator and mirror it).
- *Brightness* slider (backlight, via brightnessctl).
- *Keyboard-backlight* brightness slider (brightnessctl on the kbd_backlight class).
- *Mouse* enable/disable toggle — shown only when a mouse is connected.
- *Trackpad* enable/disable toggle — shown only when a trackpad is connected (mirror =toggle-touchpad= / =touchpad-auto=).
- *Idle inhibitor* (the =custom/idle= module that replaced the built-in =idle_inhibitor= 2026-06-24 — toggles the hypridle daemon, state-synced icon).
- *Airplane mode* (the existing =airplane-mode= toggle; laptop-only).

The conditional rows (mouse, trackpad, airplane) appear only when their hardware/context applies — reuse the laptop/device detection the airplane and touchpad indicators already do.

Design / open questions (propose before building):
- Panel tech: sliders need a real toolkit (waybar can't host a slider), so a GTK4 + gtk4-layer-shell app like pocketbook is the likely shape.
- Which existing standalone bar modules (dim, touchpad, airplane, idle_inhibitor) collapse INTO this panel vs. stay on the bar as quick-access indicators. Craig's call.

Implementation notes: a small GTK layer-shell app (mirror pocketbook's structure: src-layout Python package, pytest, Makefile) talking to brightnessctl / hyprctl / the touchpad + airplane helpers. Lives in the dotfiles repo or in-tree like pocketbook. TDD the backing toggle/slider logic. Sizable — worth a design doc first.

** TODO [#B] Local offline LLM runtime + per-host model cache :tooling:llm:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-06
:END:
Add a local-LLM provisioning track so machines can run an offline coding agent when there's no network. Install =llama.cpp= (CPU + Vulkan where practical) and prefetch per-host model files while network is available; expose OpenAI-compat local endpoints (=127.0.0.1:8081= coding, =:8082= general; =:11434= reserved for =ollama= if used). Per the rulesets generic-agent-runtime design pass — rulesets becomes runtime-neutral and owns the runtime manifests + project instructions; archsetup owns machine provisioning + the per-machine model inventory. Source: handoff from rulesets 2026-05-28 ([[file:assets/outbox/2026-05-28-from-rulesets-local-llm-install.org][outbox copy]]).

Per-host model targets (from the handoff):
- *ratio* (Strix Halo, 128 GiB) — Qwen3-Coder-30B Q6_K (default) + Q4_K_M (compat) + Qwen3-Next-80B Q4_K_M (long-context fallback).
- *velox* (i7-1370P, 64 GiB iGPU) — Qwen3-Coder-30B Q4_K_M + an 8B fallback for low-latency triage.

Install behavior: prefetch idempotent (skip if file exists, match size/hash); download failure must NOT fail the install — surface a clear "local LLM support incomplete" follow-up instead. Ship a smoke-test command (boot endpoint + short prompt).

Decisions to resolve before code:
*** TODO Decide model cache location: per-user vs system-wide
Handoff lists both =~/.local/share/llm/models= (per-user) and =/srv/models/llm= (system-wide). Per-user matches the existing archsetup user-config style and avoids root ownership of large model files. System-wide matches the "machine-local model inventory" phrasing and shares cache across users on multi-user boxes (not the case here — single user per machine). Pick one as the default; the other stays available via =LLM_MODEL_CACHE=.
*** TODO Decide whether =ollama= ships by default or is opt-in
Handoff calls =ollama= "optional". Likely shape: =llama.cpp= is the only mandatory runtime; =ollama= behind =INSTALL_OLLAMA= (default no) for users who prefer its model-manager API. Confirm.
*** TODO Define config keys for the LLM block in =archsetup.conf.example=
Likely: =INSTALL_LOCAL_LLM= (default yes), =LLM_RUNTIME= (=llama.cpp= / =ollama=), =LLM_MODEL_CACHE= (path), =LLM_MODELS= (space-separated, or empty → per-host autodetect). Lock names + defaults before writing install code.
*** TODO Decide per-host model selection: auto-detect by =uname -n= vs explicit =LLM_MODELS=
Auto-detect against a known-host table (ratio → Q6_K + 80B, velox → Q4_K_M + 8B) is simple for current machines but brittle for any new host (silently picks no models). Explicit =LLM_MODELS= per machine in =archsetup.conf= is more verbose but never surprises. Pick the default; the other stays available.
*** TODO Decide network-down behavior for model prefetch
Three shapes: (a) emit =error_warn= and write =/var/lib/archsetup/state/llm-models-pending= for inspection; (b) install a one-shot systemd unit that retries on next boot with network; (c) just log and forget — user re-runs the prefetch helper manually when network returns.

Implementation work (gated on the decisions above):
*** TODO Install =llama.cpp= with CPU + Vulkan backend where supported
Add to the appropriate install section in =archsetup= (=llama.cpp= / =llama.cpp-vulkan= in AUR). Decide CPU-only vs Vulkan per host from the hardware detection already used for GPU drivers.
*** TODO Install =ollama= behind config flag (if Decision 2 = opt-in)
Add =ollama= package install gated on =INSTALL_OLLAMA=yes=.
*** TODO Configure shared model cache + OpenAI-compat local endpoints
Create =$LLM_MODEL_CACHE= with the right ownership; configure llama.cpp (and ollama if installed) to serve =127.0.0.1:8081= (coding) and =:8082= (general). Likely systemd user units; decide launcher pattern when implementing.
*** TODO Prefetch per-host models (idempotent, non-fatal on network failure)
Download the per-host model set (from Decision 4) into the cache; skip files that exist with matching size/hash. On failure, fall back per Decision 5. Models from HuggingFace GGUF mirrors (URLs locked at implementation time).
*** TODO Ship a local-LLM smoke-test command
Boot the configured endpoint and send a short prompt; surface success/failure + timing. Useful as both a post-install check and a triage tool when something later breaks. Likely =scripts/llm-smoke-test.sh=; runs at end of install if =INSTALL_LOCAL_LLM=yes=.

Acceptance: fresh VM install of the ratio profile reaches an endpoint on =:8081= that answers a smoke prompt; velox profile gets Q4_K_M + 8B and answers a prompt within reasonable laptop latency; network-down install completes successfully with the pending-models warning surfaced.

** TODO [#C] Voice dictation / speech-to-text input :feature:tooling:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-06
:END:
Push-to-talk dictation that types transcribed speech into the focused Wayland window — usable at any text field, including the Claude Code terminal prompt and Emacs buffers. Claude Code has no built-in voice input; dictation has to happen at the OS level and inject text. Raised 2026-07-03.

Tool choice is the open decision (needs Craig): =nerd-dictation= (Vosk, lighter, lower accuracy) vs a =whisper.cpp=-based daemon (heavier, higher accuracy, optional GPU). Wayland typing backend is =wtype= or =ydotool=. Scope once chosen: install + model download, a push-to-talk keybind (Hyprland), and an autostart entry; fold into archsetup so it lands on both daily drivers. Consider an Emacs-native path (=whisper.el=) as a complement for in-buffer dictation.

** TODO [#B] Review post-archsetup laptop setup steps (velox 2026-04-10)
:PROPERTIES:
:LAST_REVIEWED: 2026-07-04
:END:
Items discovered during velox setup that needed manual intervention after archsetup.
Decide which should be automated in archsetup vs documented as post-install steps.

*** TODO Review: rfkill soft blocks bluetooth and wifi at boot
Both bluetooth and wifi were soft-blocked by rfkill. Fix was ~rfkill unblock bluetooth/wifi~.
~systemd-rfkill~ persists state, so unblocking once should stick, but new installs may default to blocked.
Consider: add ~rfkill unblock all~ to archsetup post-install or a firstboot script.

*** 2026-07-04 Sat @ 11:48:24 -0500 Automated /efi restrictive mount permissions in fstab generation
archsetup:2827-2836 now rewrites the /efi fstab line to =fmask=0177,dmask=0077= (idempotent), so fresh installs no longer land the world-accessible =fmask=0022,dmask=0022= default. Confirmed via the 2026-07-04 task audit. (Original velox note: default vfat mount had =fmask=0022,dmask=0022=, hand-fixed to restrictive; bootctl warned about a world-accessible random-seed file.)

*** TODO Review: tmpfs layered over ZFS /tmp causing systemd-tmpfiles failures
~systemd-tmpfiles-clean.service~ failed repeatedly with "Protocol driver not attached".
Root cause: systemd's ~tmp.mount~ (tmpfs) mounted before ZFS's ~/tmp~ dataset, creating a stale layer.
Fix: ~systemctl mask tmp.mount~. Consider: mask tmp.mount in archsetup when ZFS is used.

*** TODO Review: intel-ucode not installed
CPU running old microcode. Installed ~intel-ucode~ and rebuilt initramfs.
Consider: add intel-ucode (or amd-ucode) to archsetup package list based on CPU vendor.

*** 2026-07-04 Sat @ 11:48:24 -0500 Automated syncthing user-service enable in archsetup
archsetup:2263-2271 now installs syncthing and enables the user service (via symlink), so fresh installs no longer leave it installed-but-disabled. Confirmed via the 2026-07-04 task audit. (Original velox note: package installed but service not enabled; hand-fixed with =systemctl enable --now syncthing@cjennings=.)

*** TODO Review: awww-daemon crashes at boot (coredump)
Wallpaper daemon crashed with abort() shortly after boot. Hyprland also coredumped at same time.
May be a race condition. Restarting awww-daemon fixed it. Monitor for recurrence.

*** TODO Review: touchpad-indicator missing (X11 only, no Wayland equivalent)
Old ~touchpad-indicator-git~ was X11-only and removed as broken.
Created ~touchpad-auto~ (auto-disable touchpad when mouse connected) and ~toggle-touchpad~ scripts.
~touchpad-auto~ watches Hyprland socket for mouseadded/mouseremoved/configreloaded events.
Device name ~pixa3854:00-093a:0274-touchpad~ is hardcoded — will differ on other machines.
Added to exec-once and $mod+F9 keybinding.
Consider: add scripts to stowed dotfiles, make touchpad device name auto-detected.

*** TODO Review: Bluetooth mouse pairing is manual post-install
Paired Logi M650 via ~bluetoothctl scan on~, ~pair~, ~trust~.
This is inherently interactive (scan, select device, pair, trust).
Consider: document as post-install step. No automation possible.

*** 2026-05-26 Tue @ 13:32:31 -0500 pocketbook install concern moot — pulled from publication, folded in-tree
Resolved by removing pocketbook from archsetup's provisioning entirely. It's nowhere near ready, so the github mirror + cjennings.net repo were deleted and the project was folded into the archsetup tree at =pocketbook/=. Dropped the =gtk4-layer-shell= dep + =pip_install= from =archsetup= and the clone from =scripts/post-install.sh=. No fresh install pulls pocketbook now, so "not installed on velox" no longer applies. Re-wiring the install is tracked in the new pocketbook development backlog.

*** TODO Review: Tailscale needs login after install
~tailscaled~ service was enabled but needed ~tailscale up~ for interactive auth.
Old machine entry needed cleanup in admin console.
Consider: document as post-install step.

*** TODO Review: docs/ directories need manual sync from existing machine
docs/ dirs (gitignored) for ~/code and ~/projects repos needed scp/rsync from ratio.
Same for ~/.emacs.d/docs/. Not in git, so not available after clone.
Consider: document as post-install step or create a sync script.

** TODO [#B] Test + CI infrastructure :test:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-13
:END:
Umbrella for the test-harness and CI-automation buildout. Consolidated from the 2026-06-28 task audit: these were scattered top-level tasks circling one effort, re-homed as children so the work reads as a unit. Each child ships independently and keeps the priority it carried before. No CI runner exists yet, so the CI/CD-pipeline child gates several of the others.

*** TODO [#B] Build CI/CD pipeline that runs archsetup on every commit
:PROPERTIES:
:LAST_REVIEWED: 2026-06-13
:END:
Core automation infrastructure - enables continuous validation
*** TODO [#B] Generate recovery scripts from test failures
:PROPERTIES:
:LAST_REVIEWED: 2026-06-13
:END:
Auto-create post-install fix scripts for failed packages - makes failures actionable
*** TODO [#B] Establish monthly review workflow
:PROPERTIES:
:LAST_REVIEWED: 2026-06-13
:END:
The diff engine now exists (=scripts/package-inventory= / =make package-diff=), so what remains here is the cadence, not the tooling: a scheduled prompt to run the diff and act on it. Subtasks 1-2 are the recurring human judgment the engine feeds; subtask 3 is the automation to schedule it.
**** TODO [#B] For packages in archsetup but not on system: determine if still needed
**** TODO [#B] For packages on system but not in archsetup: decide add or remove
**** TODO [#B] Schedule monthly package diff review
*** TODO [#B] Set up automated test schedule
:PROPERTIES:
:LAST_REVIEWED: 2026-06-28
:END:
Weekly full run to catch deprecated packages even without commits
*** TODO [#B] Implement manual test trigger capability
:PROPERTIES:
:LAST_REVIEWED: 2026-06-28
:END:
Allow on-demand test runs when automation is toggled off
*** TODO [#B] Create test results dashboard/reporting
:PROPERTIES:
:LAST_REVIEWED: 2026-06-28
:END:
Make test outcomes visible and actionable
*** TODO [#B] Block merges to main if tests fail
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Enforce quality gate - broken changes don't enter main branch
*** TODO [#B] Add network failure testing to test suite
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Simulate network disconnect mid-install to verify resilience
*** TODO [#B] Keep VM base images up to date
:PROPERTIES:
:LAST_REVIEWED: 2026-06-28
:END:
Regular updates to the Arch base VM image (qemu, built by =create-base-vm.sh=) with a review process and schedule. The harness is VM/qemu-based, not containers.
*** TODO [#B] Persist test logs for historical analysis
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Archive logs with review process and schedule to identify failure patterns and trends
*** TODO [#B] Implement automated deprecation detection
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Parse package warnings and repo metadata to catch upcoming deprecations proactively
*** TODO [#B] Monitor and optimize test execution time
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Keep test runs performant as installs and post-install tests grow (target < 2 hours)
*** TODO [#B] Set up alerts for deprecated packages
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Proactive monitoring integrated with testing
*** TODO [#B] Fix VM cloning machine-ID conflicts for parallel testing
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Currently using snapshot-based testing which works but limits to sequential test runs
Cloned VMs fail to get DHCP/network even with machine-ID manipulation (truncate/remove)
Root cause: Truncating /etc/machine-id breaks systemd/NetworkManager startup
Need to investigate proper machine-ID regeneration that doesn't break networking
Would enable parallel test execution in CI/CD
Priority C because snapshot-based testing meets current needs

** TODO [#C] Fix install errors surfaced by the 2026-05-11 VM test run
:PROPERTIES:
:LAST_REVIEWED: 2026-07-06
:END:
*** 2026-06-28 Sun @ 13:29:29 -0400 Audit reconcile: 2026-06-28 btrfs+zfs runs reproduce the same residual set
Newer full runs landed since the 2026-06-11 reconcile below: the 2026-06-25 zfs run (Testinfra 96/0) and the 2026-06-28 btrfs+zfs runs (97/0, "zero attributed issues"). The residual four were NOT fixed and reproduce unchanged: =enabling firewall= (archsetup:1496-1498, carries a VM-kernel note), =enabling gamemode for user= (archsetup:2221, non-critical), and =tidaler (AUR)=. Zero archsetup-attributed Testinfra issues across both profiles confirms these are environment / non-critical, not archsetup bugs. Bare-metal confirmation of the firewall pair is still the open thread.

*** 2026-06-15 Mon @ 23:53:21 -0500 Audit reconcile: latest VM run (2026-06-11) confirms the surviving error set
The most recent VM run (=test-results/20260611-113904/=) carries four error-summary entries: =enabling firewall= + =verifying firewall is active= (the iptables/nf_tables "Could not fetch rule set generation id" pair, still unconfirmed on bare metal), =enabling gamemode for user= (non-critical), and =tidaler (AUR)=. The earlier fontconfig/dconf fixes held — none reappear. So the count is down from the 7→6 anchor below to four, all of them the known-residual items already itemized.
Errors logged during the VM install. Status as of the 2026-05-11 18:36 run (=test-results/20260511-183643/archsetup-output.log=) after the =48c9439= fontconfig/dconf fix: 7 → 6.
- refreshing font cache — RESOLVED in =48c9439= (now installs =fontconfig= before calling =fc-cache=).
- configuring GTK file chooser — RESOLVED in =ecab29f= (switched to a system-wide dconf db at =/etc/dconf/db/site.d/=; needs no session bus during install).
- configuring GNOME interface settings in dconf — RESOLVED in =ecab29f= (same fix as the GTK file chooser above).
- enabling firewall — exit 1: =iptables v1.8.13 (nf_tables): Could not fetch rule set generation id: Invalid argument=. Still present in the 18:36 run; likely a VM-kernel/nf_tables artifact — confirm on bare metal before treating as an archsetup bug.
- verifying firewall is active — exit 1 (follow-on from the firewall-enable error).
- enabling gamemode for user — exit 1 → step "gaming" FAILED — non-critical.
- tidaler (AUR) — logged in the error summary with exit code 0 (odd; logging quirk or transient AUR build noise?).
Also seen in the 18:36 run's log-diff (post-install systemd noise, probably VM-environment): =pam_systemd … CreateSession failed= / =logind: Failed to start session scope … Permission denied=, and =Failed to start Proton VPN Daemon= (no VPN config in the test VM).

*** 2026-05-19 Tue @ 13:18:56 -0500 Fixed AUR exit-0 logging bug at the root
Root cause was in =retry_install=: =last_exit_code=$?= ran AFTER =if eval ...; then return 0; fi=. Bash defines an if-compound's exit status as zero when no condition tested true, so a failing eval's exit code got overwritten with 0 before reaching =error_warn=. Fix in =8221c54=: capture =$?= from =eval= directly into a local var, then compare against the captured value in the if. VM-verified in =test-results/20260519-115318/=: =mkinitcpio-firmware (AUR)= and =tidaler (AUR)= now report =error code: 1= (yay's actual exit) instead of the misleading =error code: 0=. The same packages still appear in the summary because yay returns non-zero when sub-deps fail to build (e.g. =aic94xx-firmware=), but the codes are accurate now. If the underlying sub-dep failures stay noisy, that's a separate concern — open a new task.

*** 2026-05-16 Sat @ 09:00:41 -0500 AI Response: Surfaced the expanded AUR-exit-0 pattern
2026-05-16 07:40 VM run passed (52/0/5) with the same warning profile as the 2026-05-11 18:36 run. Error count went 7 → 13: 5 fixed/unchanged, +5 new AUR-exit-0 entries (broadens the existing tidaler item into the dedicated =[#B]= subtask above), +1 genuinely new error in =setting up emacs configuration files= (=git pull= ran in =~/.emacs.d= which existed from stow but had no =.git=). Patched =archsetup:1932-1945= with a three-branch check: clone if missing/empty, pull if =.git= exists, =git init=/=fetch=/=checkout= in place if the dir came from stow.

*** 2026-05-19 Tue @ 01:25:26 -0500 Verified the b9907c7 emacs-stow fix end-to-end
=make test= 21:44 → 22:29 (42 min), =test-results/20260518-214516/=. 52/0/5, =ArchSetup Exit Code: 0=. The third-branch path fired correctly — install log =archsetup-2026-05-18-21-45-46.log:14358-14365= shows =From https://git.cjennings.net/dotemacs= → =[new branch] main -> origin/main= → =Reset branch 'main'= → =branch 'main' set up to track 'origin/main'=. No exit-128, no =fatal: not a git repository=. Error Summary down to 7 (was 13 on 2026-05-16); the emacs entry is gone. AUR exit-0 logging triggered for 2 packages this run (mkinitcpio-firmware, tidaler) vs 6 on 2026-05-16 — same bug class, fewer triggers, still tracked under =[#B] AUR exit-0 logged as error=. Issue Attribution: 1 ARCHSETUP entry (Proton VPN Daemon failed — known VM-no-VPN-config artifact). Cleanup ran clean via the normal path.

** TODO [#B] Review undeclared ratio packages for installer inclusion :chore:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Triggered by the 2026-06-14 =make package-diff= run on ratio: 62 packages are installed but not declared in archsetup. Stripped of the structural buckets — pacstrap base/boot/kernel (base, linux*, grub, efibootmgr, sudo, btrfs-progs, fwupd, logrotate, ex-vi-compat, linux-lts-strix, zram-generator), the =make deps= VM set (qemu-full, virt-manager, virt-viewer, libguestfs, bridge-utils, dnsmasq, archiso), and the yay bootstrap — these 40 remain. Check the ones to add to the installer, then rerun =make package-diff= to confirm they clear.

Evidence report (2026-07-14, count now 64): [[file:docs/design/2026-07-14-undeclared-packages-evidence.org][docs/design/2026-07-14-undeclared-packages-evidence.org]] (archsetup 78081e4) — description, requirer, and install date per package, bucketed candidates (30) / dependency-pulled (6) / orphan libraries (7) / structural (21). The include/ignore pass reads the candidates bucket; the checklist below stays until Craig walks it.

Some entries are libraries likely pulled in as dependencies (blas-openblas, openblas, eigen, tk, lib32-openal, pkcs11-helper, gtk4-layer-shell, webkit2gtk, sane, freerdp, rust-bindgen) — check those only if you want them declared explicitly rather than left to dependency resolution.

- [ ] aws-cli-v2
- [ ] bats
- [ ] blas-openblas
- [ ] drawio-desktop
- [ ] eigen
- [ ] emacs
- [ ] flatpak
- [ ] freerdp
- [ ] geeqie
- [ ] git-lfs
- [ ] github-cli
- [ ] gtk4-layer-shell
- [ ] hugo
- [ ] imv
- [ ] lc0
- [ ] lc0-network-sm
- [ ] ledger
- [ ] lib32-openal
- [ ] libreoffice-fresh
- [ ] minidlna
- [ ] openai-codex
- [ ] openblas
- [ ] pacoloco
- [ ] pkcs11-helper
- [ ] proton-vpn-cli
- [ ] proton-vpn-daemon
- [ ] protontricks
- [ ] python-lyricsgenius
- [ ] python-pip
- [ ] python-pipx
- [ ] python-sphinx
- [ ] rust-bindgen
- [ ] sane
- [ ] shortwave
- [ ] spotify-launcher
- [ ] tidal-dl-ng
- [ ] tk
- [ ] typescript-language-server
- [ ] webkit2gtk
- [ ] whisper.cpp

** TODO [#B] Security hardening + audit :security:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-14
:END:
Umbrella for the security-hardening and audit effort. Consolidated from the 2026-06-28 task audit, re-homing the scattered security tasks as children so the work reads as a unit. Each child ships independently and keeps its prior priority.

*** TODO [#B] Test security + functionality together
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
**** TODO [#B] Verify no unexpected open ports or services
*** TODO [#B] Security audit tooling
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
**** TODO [#B] Implement port scanning check
**** TODO [#B] Create security posture verification script
**** TODO [#B] Set up intrusion detection monitoring
*** TODO [#B] Document threat model and mitigations within 6 months
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Identify attack vectors, what's mitigated, what remains
*** TODO [#B] Complete security education within 3 months
:PROPERTIES:
:LAST_REVIEWED: 2026-06-24
:END:
Read recommended resources to make informed security decisions (see metrics for Claude suggestions)
*** TODO [#B] Create security checklist for cafe/public wifi scenarios
:PROPERTIES:
:LAST_REVIEWED: 2026-05-21
:END:
Practical guidelines for working in public spaces

** TODO [#B] Ensure sleep/suspend works on laptops
:PROPERTIES:
:LAST_REVIEWED: 2026-07-06
:END:
Critical functionality for laptop use - current battery drain unacceptable
*NOTE:* This applies to Framework Laptop (velox), not Framework Desktop (ratio)
Add kernel parameter: ~rtc_cmos.use_acpi_alarm=1~ (will become systemd default)
Consider: ~acpi_mask_gpe=0x1A~ for battery drain, suspend-then-hibernate config
See Framework community notes on logind.conf and sleep.conf settings

** TODO [#C] Re-check python-lyricsgenius --skipinteg workaround :chore:solo:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
archsetup installs =python-lyricsgenius= with =--mflags --skipinteg=, skipping makepkg integrity + PGP checks — a workaround originally for an expired-signature issue upstream (surfaced by the 2026-06-23 --noconfirm audit). Periodically test whether the cause has cleared: if a plain =aur_install python-lyricsgenius= builds without complaint, drop the =--skipinteg= workaround. Removal needs a real AUR build to confirm, so it isn't a blind change.

*** 2026-07-14 Tue @ 01:40:48 -0500 Rechecked: still needed, same cause
Fresh AUR clone, =makepkg --verifysource= on 3.7.0-1 (PKGBUILD unchanged): tarball passes, =LICENSE.txt= still FAILS its b2sum — the github-master pin, structural as diagnosed 2026-06-24. =--skipinteg= stays.

*** 2026-07-02 Thu @ 05:08:37 -0400 Rechecked: still needed, same cause
=makepkg --verifysource= on 3.7.0-1: tarball passes, =LICENSE.txt= still FAILS
its b2sum — the PKGBUILD still pins a hash for a file fetched from github
master. Structural, as diagnosed 2026-06-24; =--skipinteg= stays.

*** 2026-06-24 Wed @ 17:55:34 -0400 Rechecked: still needed, but the cause changed
Ran =makepkg --verifysource= on the current AUR PKGBUILD (3.7.0-1). The package tarball =lyricsgenius-3.7.0.tar.gz= now passes its b2sum — the original expired-PGP-signature problem is gone (the PKGBUILD no longer carries any =validpgpkeys=). But integrity still FAILS, on a different file: =LICENSE.txt=, which the PKGBUILD fetches from the project's github master and pins a b2sum for. github master is a moving target, so that b2sum drifts and =--skipinteg= is still required. This is structural (not a transient upstream fix away), so it likely won't clear until the maintainer pins the LICENSE to a tagged release. Updated the archsetup comment to the real cause. Keep rechecking, but lower expectations of it clearing.

** TODO [#C] Review theme config architecture for dunst/fuzzel
:PROPERTIES:
:LAST_REVIEWED: 2026-07-06
:END:
The active dunst config is stowed from dotfiles/common/ but theme templates
live in dotfiles/hyprland/.config/themes/. set-theme copies the templates to
the stowed locations at runtime, so edits to the common file get overwritten
on theme switch. This split between stowed configs and theme templates is
error-prone — changes must be made in both places. Consider:
- Having set-theme be the single source of truth (remove common dunstrc from stow)
- Or symlinking the stowed config to a theme-managed location
- Same situation applies to fuzzel.ini
The goal is a single place to edit each config, not two.

** TODO [#C] Review current tool pain points annually
:PROPERTIES:
:LAST_REVIEWED: 2026-07-06
:END:
Once-yearly systematic inventory of known deficiencies and friction points in current toolset

** TODO [#D] Consider Customizing Hyprland Animations
Current: windows pop in, scratchpads slide from bottom.

Customizable animations:
- windows / windowsOut / windowsMove - window open/close/move
- fade - opacity changes
- border / borderangle - border color and gradient angle
- workspaces - workspace switching
- specialWorkspace - scratchpads (currently slidevert)
- layers - waybar, notifications, etc.

Styles: slide, slidevert, popin X%, fade
Parameters: animation = NAME, ON/OFF, SPEED, BEZIER, STYLE
Speed: lower = faster (1-10 typical)

Example tweaks:
#+begin_src conf
animation = windows, 1, 2, myBezier, popin 80%
animation = workspaces, 1, 4, default, slide
animation = fade, 1, 2, default
animation = layers, 1, 2, default, fade
#+end_src

** TODO [#D] Parse and improve AUR error reporting
Parse yay errors and provide specific, actionable fixes instead of generic error messages

** TODO [#D] Improve progress indicators throughout install
Enhance existing indicators to show what's happening in real-time

** TODO [#B] Manual testing and validation :test:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Craig's standing checklist of everything that isn't agent-verifiable. Each child is one test in the =verification.md= shape (title, what we're verifying, steps, Expected). A child that fails gets its actual behavior written under it and is promoted to a top-level TODO. 44 checks pending as of the 2026-07-09 audit.

Priority and type tag added by that audit: the task carried neither, which kept the project's largest live container out of the agenda entirely.

*** Net panel: sticky error toast dismisses on the next click
What we're verifying: the stuck enterprise-error banner (dotfiles a157bed) now clears when you click anywhere else in the panel — the fix is a pointer-level gesture that AT-SPI can't drive.
- Open the net panel from the bar.
- Click an enterprise (802.1X) network in the NETWORKS list (Cox Mobile in the usual scan) so the red "Enterprise (802.1X) — join or edit it in nmtui/nmcli." banner appears.
- Click anywhere else in the panel: another network row, a meter card, or empty plate.
Expected: the red banner disappears on that click. A routine (non-red) toast still fades on its own 4s timer as before.

*** maint privilege degrade: PROMPT path on a real tty
What we're verifying: with sudo not passwordless, a maint fix on a terminal prompts for the password (after the wall note) instead of hard-failing. Safe on ratio/velox — run from a real terminal; PANELKIT_SUDO=false makes the model treat the box as non-passwordless without touching sudoers, while the fix still fires through real (prompting) sudo.
#+begin_src sh :results output
PANELKIT_SUDO=false maint fix cache_clean
#+end_src
- Type your password if sudo asks.
Expected: a dim note line "no passwordless sudo — sudo will ask for your password on this terminal", then the RUN line shows plain "sudo paccache -r" (no -n) and the step lands OK with re-probe notes after. On ratio/velox the NOPASSWD grant means sudo won't actually prompt (only the seam is faked); the real password prompt is only observable on a box without the grant — a VM or a fresh install pre-sudoers.

*** maint privilege degrade: GUIDE events on the GUI wall
What we're verifying: the maint panel's CLEAN UP degrades to guide instructions on the results wall (nothing executes) when the box can't elevate silently.
#+begin_src sh :results output
PANELKIT_SUDO=false setsid -f maint panel >/dev/null 2>&1
#+end_src
- In the panel, press CLEAN UP (needs an off-nominal reclaim metric; if the board is quiet, expect the "nothing to clean" note instead).
Expected: each privileged reclaim lands on the wall as a dim instruction line — "cannot elevate here (...) — run yourself: $ sudo ..." — with no RUN/OK pair for it, and nothing actually reclaimed. Close the panel afterward (it inherited the fake PANELKIT_SUDO).

*** Net doctor privileged fixes (net Phase 1) — run in a disposable VM/container
What we're verifying: the three control-plane fixes actually repair a real broken state under real sudo. These states are dangerous on a daily driver (masking NM kills the network), so run them in a throwaway VM or booted nspawn with NetworkManager installed and passwordless sudo, NOT on ratio/velox. Each: break the state, run =net doctor --fix=, confirm the repair.
**** nm-masked → unmask-nm
What we're verifying: a masked NetworkManager is unmasked and started, not met with a doomed nm-restart.
#+begin_src sh :results output
sudo systemctl mask NetworkManager
net doctor --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print(d["outcome"], d.get("next_action"))'
#+end_src
- Expected (diagnose): the verdict names the masked NM (outcome fixable, action unmask-nm), not the generic "NetworkManager isn't running".
#+begin_src sh :results output
net doctor --fix --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print("fixed=",d["fixed"],"attempts=",[a["id"] for a in d["attempts"]])'
systemctl is-enabled NetworkManager; systemctl is-active NetworkManager
#+end_src
Expected: the fix attempts unmask-nm, NetworkManager is no longer masked (is-enabled != masked) and is active.
**** rival-manager → disable-rival
What we're verifying: a rival manager active alongside NM is disabled, and the doctor names it rather than blindly resetting.
#+begin_src sh :results output
sudo systemctl enable --now dhcpcd   # a rival that fights NM for the link
net doctor --fix --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print(d["outcome"], [a["id"] for a in d["attempts"]])'
systemctl is-active dhcpcd
#+end_src
Expected: the verdict/fix is disable-rival, and dhcpcd reads inactive afterward. Note whether the link comes back on its own or needs a follow-up =net doctor --fix= (the un-chained-reset question — if it needs the reset, chain "disable-rival": ["disable-rival","reset"] in FIX_CHAINS).
**** keyfile-perms → chmod-keyfile
What we're verifying: a profile keyfile with unsafe perms is set back to 0600 root so NM stops ignoring it. Needs root (the system-connections dir is 0700 root:root), so run the doctor as root or via sudo.
- Pick an existing profile name (its keyfile is /etc/NetworkManager/system-connections/<name>.nmconnection).
#+begin_src sh :results output
sudo chmod 0644 /etc/NetworkManager/system-connections/<PROFILE>.nmconnection
sudo net doctor --fix --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print(d["outcome"], [a["id"] for a in d["attempts"]])'
sudo stat -c '%a %U' /etc/NetworkManager/system-connections/<PROFILE>.nmconnection
#+end_src
Expected: the verdict/fix is chmod-keyfile and the keyfile reads =600 root= afterward.

*** Net doctor sharpened auth verdict (net Phase 2) — needs a real WPA3 / hidden network
What we're verifying: the auth-cause classifier names the specific auth cluster (SAE / hidden / enterprise / generic) instead of a bare "password rejected", and the SAE / hidden one-line profile fixes actually run and take. These need a real network the machine can (fail to) associate with, so they can't be faked; run against an AP you control or a VM bridged to one. The classify logic is already unit-tested — this is the live half.
**** SAE cause named + auth-sae fix
What we're verifying: a WPA3-only network whose saved profile is still WPA2-PSK reports the SAE cause, and =--fix= sets key-mgmt sae so the association can complete.
- Set an AP (or its VM equivalent) to WPA3-Personal only (pure SAE, not WPA2/WPA3 transition).
- Save a profile for it as WPA2-PSK (=nmcli connection add type wifi ... wifi-sec.key-mgmt wpa-psk=), then try to connect so the association fails on auth.
#+begin_src sh :results output
net doctor --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print(d["outcome"], d.get("message"), "|", d.get("next_action"))'
#+end_src
- Expected (diagnose): outcome fixable, action auth-sae, the message names WPA3 (not just "password rejected").
#+begin_src sh :results output
net doctor --fix --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print("fixed=",d["fixed"],"attempts=",[a["id"] for a in d["attempts"]])'
nmcli -g 802-11-wireless-security.key-mgmt connection show <PROFILE>
nmcli -g 802-11-wireless-security.pmf connection show <PROFILE>
#+end_src
Expected: attempts start with auth-sae (then reset), the profile now reads key-mgmt =sae=, and pmf is accepted (the =pmf optional= keyword took — this is the one value only a live nmcli can confirm). Note whether the link comes back after the chained reset.
**** hidden SSID cause named + auth-hidden fix
What we're verifying: a hidden (non-broadcast) network whose profile lacks the hidden flag reports the hidden cause, and =--fix= sets 802-11-wireless.hidden yes so NM probes for it.
- Set the AP SSID to non-broadcast (hidden). Save a profile for it without =802-11-wireless.hidden yes=, then try to connect.
#+begin_src sh :results output
net doctor --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print(d["outcome"], d.get("message"))'
net doctor --fix --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print([a["id"] for a in d["attempts"]])'
nmcli -g 802-11-wireless.hidden connection show <PROFILE>
#+end_src
Expected: diagnose names the hidden cause (outcome fixable, action auth-hidden); the fix attempts auth-hidden; the profile now reads hidden =yes=.
**** enterprise / generic stay terminal + named
What we're verifying: an enterprise (802.1X) auth failure stays needs-user-action and names the missing cert rather than offering a bogus fix; a plain wrong-password stays the generic "rejected" message.
#+begin_src sh :results output
# On an enterprise network with a broken/absent CA cert, or a WPA2 network with a wrong saved password:
net doctor --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print(d["outcome"], "|", d.get("next_action"))'
#+end_src
Expected: enterprise → outcome needs-user-action, next_action names the CA certificate / identity (no auth-sae/auth-hidden action); wrong password → needs-user-action with the "re-enter the password" message.

*** Bt doctor persistent-power fix (bt Phase 2) — proves out only across a real reboot
What we're verifying: a deliberately boot-disabled adapter reports =powered-off-persistent= (not the transient =power-on=), =bt doctor --fix= clears the cause and powers it on, and the adapter comes up on its own after a reboot. The reboot-survival is the whole point, so this can't be faked; run on a machine (or VM) you can reboot. Back up =/etc/bluetooth/main.conf= first.
**** AutoEnable=false → persistent verdict → fix → survives reboot
What we're verifying: the full loop — break persistence, confirm the verdict, fix, reboot, confirm it stuck.
#+begin_src sh :results output
sudo cp /etc/bluetooth/main.conf /etc/bluetooth/main.conf.bak
# Force the deliberate disable and power the adapter off:
printf '[Policy]\nAutoEnable=false\n' | sudo tee -a /etc/bluetooth/main.conf
bluetoothctl power off
bt doctor --json | python3 -c 'import sys,json; d=json.load(sys.stdin); p=[s for s in d["steps"] if s["id"]=="powered"][0]; print(p["code"], "|", p["evidence"])'
#+end_src
- Expected (diagnose): the powered step reads code =powered-off-persistent= and the evidence names AutoEnable=false (not the plain "adapter is powered off").
#+begin_src sh :results output
bt doctor --fix --json | python3 -c 'import sys,json; d=json.load(sys.stdin); print("attempts=",[a["id"] for a in d["attempts"]])'
grep -i autoenable /etc/bluetooth/main.conf
bluetoothctl show | grep -i powered
#+end_src
- Expected (fix): attempts include persist-power; main.conf now reads =AutoEnable=true=; the adapter is powered on now.
- Reboot the machine.
#+begin_src sh :results output
bluetoothctl show | grep -i powered   # after the reboot, before touching anything
#+end_src
Expected: the adapter is powered on straight out of the reboot (AutoEnable=true held). Restore the backup if desired: =sudo mv /etc/bluetooth/main.conf.bak /etc/bluetooth/main.conf=.
**** service-disabled and TLP causes (optional variants)
What we're verifying: the other two persistent causes are named and fixed the same way.
- Service variant: =sudo systemctl disable bluetooth= (leave it running this boot), power the adapter off, then =bt doctor --json= should read =powered-off-persistent=; =--fix= should re-enable it (check =systemctl is-enabled bluetooth=).
- TLP variant (only if TLP is installed): add =bluetooth= to =DEVICES_TO_DISABLE_ON_STARTUP= in =/etc/tlp.conf=, power off, diagnose (persistent), =--fix=, confirm bluetooth is dropped from the list.
Expected: each variant names the persistent verdict and the fix clears exactly that cause, leaving the others untouched.

*** Maintenance console — in-person checklist
Re-homed here by the 2026-07-09 audit: this was a second top-level "Manual testing and validation" parent. One parent, per verification.md. Priority and the :test: tag now live on the parent.
Promoted from the build parent when it closed 2026-07-08 — Craig's checklist, runs once in person. Collects everything not agent-verifiable; populate per verification.md as phases land. Known so far: panel look-and-feel vs the E5 prototype (Craig's eyeball); arm-press wording reads right at the moment of use; results-wall readability during a real doctor run; a real UPDATE through the armed guard (and the TTY path once); REBOOT offer after an update; velox in-person glyph check (battery %, charging glyph, low-charge red); SET 80% charge limit sticks across a charge cycle; KILL on a real memory hog.
Phase 8 additions: subpanel scroll position survives an arm press on a long digest (MAINT_PANEL_FIXTURE=bad, PACKAGES band, scroll into the 51-orphan list, press REMOVE — the list should stay put and the key read REMOVE?); a KEEP on a real orphan dims it into the kept section and UNKEEP restores it; MERGE on a needs-merge pacnew opens pacdiff in a foot terminal.
Phase 9a additions: MARK KNOWN on a real journal group (ratio's wpa_supplicant bgscan spam is the standing candidate) — arm shows the identifier + sample wording, fire writes curation.toml, the next re-scan drops the group into KNOWN NOISE with "marked <date>", and UNMARK restores it to SIGNAL; OPEN JOURNAL opens journalctl -p 3 -b in a foot terminal; a failed unit's RESTART on a real unit (systemctl restart via the armed key) lands the re-probe on the act line.
Phase 9b additions: MARK EXPECTED on ratio's real unexpected listeners (postgres :5432, docker :8080 — the standing curation material) dims them into the expected list and UNMARK restores them; a CPU-mode press (e.g. BAL·PWR) writes the EPP hint and the row re-reads it; KILL renders disabled (insensitive, tooltip) on a session-critical name in top memory; on velox the battery row shows NO SET 80% key (cros_ec knob absent — correct withholding, not a bug).
Phase 11 additions: glyph left-click opens the console and a second click closes it; right-click still toggles the btop scratchpad; the glyph color matches the board's worst lever-less diagnostic (open the console and check the watch items against the glyph tint); after fixing/curating a diagnostic, the glyph follows within one 30-min scan (or immediately after a manual =maint scan=); on velox in person — battery % + charging icon in the bar, collapsed bar keeps the battery glyph, low-charge red (drain below 15% or temporarily raise battery_low_pct in the TOML); after velox's reboot confirm systemctl --user list-timers shows both maint timers scheduled.
Phase 10 additions: results-wall readability during a real doctor run (press CLEAN UP on the live board — every step should appear as a running line the moment it starts, resolve in place with the result, and the re-probed truth should follow; the wall should feel like watching the run, not reading a report afterward); COPY puts the same lines the CLI wall prints on the clipboard, paste somewhere to check; HIDE collapses the wall to its header and SHOW restores it with the scroll position at the bottom; REVIEW & FIX reads right at the moment of use (each row: what's wrong, the maint fix usage, and either a FIX key or "per-item — on its subpanel"); RECLAIM on a degraded disk row arms with the first step + "(+N more steps)" and the fire streams every step to the wall individually; a guard-tripped UPDATE fired from the REVIEW roster re-arms with the override wording and the second press actually runs (the token fix — verify the loop terminates).
Phase 11b additions: the fidelity eyeball — open the live console beside the settled E5 prototype (headless-Chrome render or the browser) and judge whether the board finally reads as the same instrument: big-number typography, bar fills with the gold threshold tick on the cache card, the scrub/keyring/topgrade rings, status chips, the lit CPU-mode segment, selector count chips + severity left borders, right-aligned attention/ok/fixable/watch header; press an EPP segment on the card (e.g. BAL·PWR) and confirm it arms then fires exactly like the old digest control; on the bad fixture (MAINT_PANEL_FIXTURE=bad maint panel) check a long readout (the BACKUPS card) ellipsizes with a tooltip carrying the full value.
**** 2026-07-08 Wed @ 06:31:00 -0500 Fixed TOPGRADE remedy failure found in testing (dotfiles 9b5384f)
Craig's first testing find: TOPGRADE exited 2 on every launch — topgrade 17.6.1 names the git step git_repos and the remedy passed --disable git, rejected by clap before any step ran. Fixed red-first (test pins git_repos), full make test green, dry-run on the installed binary confirms the step name lands and the repo-rebasing step stays excluded. Velox pulled. RE-TEST: press TOPGRADE again (it's guard-armed on ratio's pending mesa set — TTY is still the safe path for the mesa run).

*** CPU mode survives a reboot
What we're verifying: =maint-epp-restore.service= replays the CPU mode the pill last set. The kernel resets EPP to the driver default on every boot, so this is the only check that can prove the fix; nothing an agent runs before a reboot distinguishes "it works" from "nothing was reset yet."
- Open the maintenance console, MEM·PWR band, and press a CPU-mode segment other than the current one (say BAL·PWR).
- Confirm the act line says the mode was remembered.
#+begin_src sh :results output
cat /sys/devices/system/cpu/cpu0/cpufreq/energy_performance_preference
grep -A1 '^\[power\]' ~/.config/maint/*.toml 2>/dev/null
#+end_src
- Reboot.
- After logging back in, run the block below.
#+begin_src sh :results output
cat /sys/devices/system/cpu/cpu0/cpufreq/energy_performance_preference
systemctl --user status maint-epp-restore.service --no-pager | head -5
#+end_src
Expected: the EPP value after the reboot is the mode you selected, not the driver default, and the unit shows =Active: inactive (dead)= with =Result=success= (a oneshot that ran and exited).

*** Audio signal metering: apply the package (precondition)
What we're verifying: the signal-metering build is live. Only a =git pull= is needed — the =audio/= package .py files are stowed and no new launcher files were added, so no restow or reboot. The shared panel CSS gained a =.lamp.dim= class (live-idle); it's in the same stowed =panel.css=.
#+begin_src sh :results output
cd ~/.dotfiles && git pull
#+end_src
Expected: the pull succeeds (commits 6054d3d..21437b4).
*** Three-state activity lamp reads live
What we're verifying: the INPUT/OUTPUT lamps show muted / live-idle / live-active from real device state, not just mute.
- Open the audio panel (bar sound glyph / Super+A). With nothing playing and the output unmuted, look at the OUTPUT lamp.
Expected: a dim (not bright) green lamp — live-idle: unmuted but no audio flowing.
- Start playing audio (music, a video).
Expected: the OUTPUT lamp brightens to a bright glowing green — live-active.
- Mute the output (click OUTPUT).
Expected: the lamp goes red — muted.
*** OUTPUT needle tracks real signal, not the fader
What we're verifying: the OUTPUT VU needle moves with actual output level, replacing the old volume-fed needle.
- With audio playing, watch the OUT · PLAY needle; then pause the audio (leave the volume fader where it is).
Expected: the needle deflects and rides the music while playing, and falls to rest when paused even though the fader hasn't moved. (The old behavior held the needle up at the fader position.)
*** INPUT needle registers the mic, including during PTT
What we're verifying: the IN · MIC needle shows real mic level and confirms PTT is capturing.
- With the mic live (INPUT unmuted), speak.
Expected: the IN · MIC needle deflects with your voice.
- Set the mic to muted, then hold the PUSH·TALK key and speak.
Expected: while held, the INPUT lamp reads live and the IN · MIC needle registers your voice; on release it rests again.
*** No parec / CPU cost when the panel is closed
What we're verifying: the meter runs only while the panel is open and leaves no orphan =parec=.
- Open the audio panel, then close it (✕, Esc, or the bar click).
#+begin_src sh :results output
pgrep -af 'parec .*@DEFAULT_(MONITOR|SOURCE)@' || echo "no parec running — clean"
#+end_src
Expected: no matching =parec= process after the panel is closed (the process groups were reaped on teardown).
*** Per-device activity lamp on each device row (option 3)
What we're verifying: each OUTPUTS/INPUTS row leads with a lamp showing whether THAT device is live — red (muted) / dim green (unmuted, idle) / bright green (carrying audio) — so you can see which specific device has sound, not just the default.
- Open the panel. Scan the leading lamp on each output/input row.
Expected: muted devices show a red lamp; unmuted devices show green.
- Play audio to a non-default output (or switch the default to a device, then play).
Expected: the row of the device actually playing shows the brightest (live-active) green; an unmuted-but-silent device shows the dimmer (live-idle) green. Judge whether the dim-vs-bright contrast is clear enough — if not, say so and the CSS is a one-line tweak.
- Note: the row lamps update on the panel's status refresh (~3s), so allow a moment after audio starts.
*** Timer redesign: apply the package + wtimer, clear stale timers (precondition)
What we're verifying: the redesigned =timer/= package + =wtimer= are live. Only a =git pull= is needed — the package .py files and =wtimer= are already stowed, and this build added no new launcher files, so no restow or reboot. Clear any pre-existing timers first so a stale-shape state file (an old pomodoro item lacking the new =cfg=) can't crash =wtimer render=.
#+begin_src sh :results output
cd ~/.dotfiles && git pull
wtimer cancel-all
#+end_src
Expected: the pull succeeds and the bar timer glyph goes idle (no active items).
*** Panel opens hero-on-top; close via the ✕, Esc, and the bar
What we're verifying: the redesigned layout renders (header, hero, CONFIGURE, queue) and every close path works.
- Click the timer module on the bar.
Expected: the panel opens floating top-right — a header faceplate, a HERO block for the primary item, the CONFIGURE strip, then a QUEUE well below.
- Click the ✕ in the header; reopen; press Esc; reopen; click the bar module again.
Expected: each of ✕, Esc, and the bar click closes the panel.
*** Timer with repeat
What we're verifying: the REPEAT toggle re-arms a timer on fire instead of dropping it.
- Pick TIMER, type =10s= (fast to observe), toggle REPEAT on, click +.
Expected: the timer appears carrying a "repeat" badge and a "timer · repeats" sub-line.
- Let it fire.
Expected: it notifies and immediately re-arms to a fresh full countdown rather than vanishing.
*** Alarm: recurring day, snooze, ringing, dismiss
What we're verifying: a recurring alarm rings (not fire-and-vanish), snooze re-arms, and dismiss re-arms a recurring one.
- Pick ALARM, type a clock time about a minute ahead (HH:MM), select today's weekday in the day picker, set SNOOZE to 1, click +.
Expected: the row shows the fire clock time and a weekday badge; the hero donut is a countdown ring to the fire time.
- Let it reach the fire time.
Expected: the hero shows RING and the bar glyph goes urgent (RING); SNOOZE and DISMISS buttons appear on the hero.
- Click SNOOZE.
Expected: ringing stops and the alarm re-arms one minute out.
- Let it ring again, click DISMISS.
Expected: ringing stops; because it is recurring it re-arms to its next matching day (a one-shot alarm would be removed instead).
*** Pomodoro: configurable cycle, auto vs await, cycle dots
What we're verifying: the config grid drives the cycle and the AUTO toggle switches between auto-advance and awaiting a start.
- Pick POMODORO, tap a named cycle (e.g. Deep) to fill the grid, set WORK S to 1 (min) for a fast run, toggle AUTO off, click + ADD CYCLE.
Expected: a pomodoro starts in work; the hero shows cycle dots and "work · cycle 1/N".
- Let the work phase fire with AUTO off.
Expected: it enters the rest phase awaiting a start — the hero shows "ready · start break" and a START BREAK button, and it does not count down until pressed.
- Press START BREAK.
Expected: the rest phase counts down; the glyph/color shifts to the break (sage) state.
*** Stopwatch: sweep dial, lap badge, promote, stop
What we're verifying: the analog sweep dial animates, the last-lap ghost badge shows, promote works, and STOP just stops (run-save deferred).
- Pick STOPWATCH, click +. If it isn't the hero, promote it (the ▲ on its queue row).
Expected: the hero donut is an analog second-hand dial sweeping once per minute (not a percentage ring); the elapsed value counts up.
- Hit LAP a couple of times (name one via the popover).
Expected: a "LAP m:ss" ghost badge appears beside the elapsed value and the sub-line shows the lap count.
- Hit STOP.
Expected: the stopwatch is removed. No =~/org/stopwatch-runs.org= entry is written — run-save is deferred to a later version.
*** Queue sort, promote/cycle, 10-item cap
- Add a 25m timer, a 5m timer, and an alarm a few minutes out.
Expected: the queue orders soonest-first; a ringing alarm jumps above everything.
- Use ‹ / › on the hero (or ▲ on a queue row) to change the primary.
Expected: the hero (bar-slot) item changes accordingly and the bar glyph follows.
- Queue items until 10 exist.
Expected: + disables with a "queue is full (10/10)" reason; running countdowns advance live (the =wtimer watch= subscription).
*** Presets: locked defaults, half-past, custom add/delete
- On TIMER, note the default chips carry no × (locked); add a custom preset via + preset, then delete it with its ×.
Expected: locked defaults can't be deleted; a custom preset adds and deletes cleanly.
- On ALARM, tap the half-past chip, click +.
Expected: an alarm is created at the next :30.
*** Bar tooltip parity
What we're verifying: the bar tooltip mirrors the redesign verbatim.
- With a pomodoro, a paused timer, and a ringing alarm active, hover the bar module.
Expected: each item lists on its own line — the pomodoro as "label  cycle/iv countdown" (no phase word), the paused one with a "(paused)" suffix, the ringing one as "RING (ringing)".
*** Audio panel: apply the new shims + configs (precondition for the tests below)
What we're verifying: the new =audio=/=audio-panel=/=waybar-audio= bin shims and the hyprland.conf + waybar config edits are live. New files need a restow (a plain =git pull= doesn't symlink them). Quit Hyprland or run from a TTY — restowing live Hyprland writes a stub hyprland.conf.
#+begin_src sh :results output
cd ~/.dotfiles && git pull
cd ~/.dotfiles && make restow hyprland
#+end_src
- Reload waybar: mod+B (or =killall -SIGUSR2 waybar=).
- Reload the Hyprland config: mod+Shift+R (or =hyprctl reload=).
Expected: =which audio audio-panel waybar-audio= resolves all three to ~/.local/bin symlinks; no stow conflicts reported.

*** Audio panel: opens and reads the live graph
What we're verifying: Super+A opens the instrument-console panel (not the old pulsemixer scratchpad) and it shows the real devices.
- Press Super+A.
Expected: the audio panel opens top-right. OUTPUTS lists every sink, INPUTS every mic (no .monitor entries), the default device in each is emphasized (cream name, gold glyph), each row shows its volume percent, and the twin OUT·PLAY / IN·MIC needles deflect. Esc closes it.

*** Audio panel: set default, volume fader, per-device mute
What we're verifying: the three row interactions drive the real graph and the panel re-reads to confirm.
- With two or more outputs present, click a non-default output row's body.
Expected: that device becomes default (gold DEF emphasis moves to it) and playback follows.
- Drag a device's fader.
Expected: the volume percent tracks the fader and the actual device volume changes (no lag, no jump to the border).
- Click a device's trailing mute glyph.
Expected: that one device mutes/unmutes; the glyph and caption reflect it; other devices unaffected.

*** Audio panel: master quick-mute (faceplate switch + mute key)
What we're verifying: the master switch and the XF86AudioMute key both mute every output at once (not just the default sink).
- Flip the faceplate master switch.
Expected: every output mutes, the state word reads MUTED, the badge shows, the faceplate glyph goes to the slashed speaker. Flip back to restore.
- With two outputs audible, press the hardware mute key (XF86AudioMute).
Expected: both outputs mute (master), not just the default. Press again to unmute all.

*** Audio panel: mic modes + push-to-talk (the meeting case)
What we're verifying: LIVE/MUTED work, and PUSH·TALK holds the mic muted except while the talk key is held — the one genuinely new capability, and the one only a live test can confirm.
- Click LIVE, then MUTED, watching the default mic row.
Expected: LIVE unmutes the mic (needle lifts), MUTED mutes it (needle red); the active mode key shows a gold lamp.
- Click PUSH·TALK. In a call app (or =audio status= in a loop), watch the mic while you hold Space, speak, then release.
Expected: mic is muted at rest, un-mutes for exactly as long as Space is held, re-mutes on release. Switching back to LIVE or MUTED disarms the hold (Space types normally again).

*** Audio panel: visual polish eyeball + bar entry point
What we're verifying: the panel looks right in the dupre instrument-console language, and the bar opens it.
- Look over the whole panel: faceplate spacing, engraved section labels, fader styling, VU needle centering, the muted-vs-audible glyph/color states.
Expected: it reads as a sibling of the net and bt panels; nothing overflows the gold border; the faders and gauges look machined, not stock GTK.
- Right-click the waybar sound module.
Expected: the panel opens (left-click still mutes, scroll still changes volume).

*** SUPERSEDED — Timer fuzzel dialogs (Escape-cancel, 12h alarm entry)
These two tests exercised the old fuzzel creation dialogs, retired when the bar's =custom/timer= on-click became =timer-panel=. The panel redesign checklist above covers the same ground: alarm entry (12h shapes, bad-input reject) is now the panel's ALARM freeform + inline validation, and the whole-flow abort is the panel's Esc/✕/bar close. Nothing to run here; kept as a pointer so the intent isn't lost.

*** Speed test streams in the panel
What we're verifying: the panel's speed test fills in as phases complete instead of dumping everything at the end (the CLI path is live-verified; this is the GTK rendering).
NOTE (2026-07-04 audit): these steps describe the OLD four-tab Blueprint panel ("Diagnostics → Network Performance"). The net panel was rebuilt as a single-screen instrument console (dotfiles 800ef60, f4e688e dropped the Blueprint pages), so the navigation below no longer matches the shipped UI — reword to the console layout before running. Don't delete; the streaming-render behavior is still worth verifying.
- Open the net panel, Diagnostics → Network Performance → Run Speedtest.
Expected: a Ping/Download/Upload checklist appears under the running row; ping lands within seconds, download mid-run, upload near the end; then the final rows (with jitter on Ping, a Packet loss row) replace it. A Tip row appears only if a rule fired, and it names the numbers that triggered it.

*** Proton VPN CLI sign-in (velox now, ratio on its trip)
What we're verifying: the proton CLI has its own account store (separate from the retired GTK app's), so the panel's proton rows can't toggle until you sign in once per machine.
- Run in a terminal: protonvpn signin <your-proton-username> (it prompts for the password). The CLI's action is =signin=, not =login=.
#+begin_src sh :results output
protonvpn info
#+end_src
Expected: Account shows your Proton username instead of 'None'. After that, protonvpn status still says Disconnected — correct, nothing auto-connects.

*** Tunnels round-trip: panel rows + bar badge (first real tunnel-owned route)
What we're verifying: the panel's Tunnels tab drives a real wireguard tunnel up and down, and the bar indicator grows the vpn badge while the tunnel owns the default route (the badge has never rendered live — every prior check ran with the wlan owning the route).
- Open the net panel (left-click the bar's net module), switch Connections to the Tunnels page.
- Confirm the rows: tailscale (up), and the three Proton configs (wg-US-TX-714, wg-US-CA-144, wg-NL-781), all down.
- Select wg-US-TX-714, press Bring Up, wait for the row to land.
Expected: the bar's net glyph gains the small vpn badge; its tooltip names the owner ("Tunnel: default route via wgpvpn (wireguard)").
- Press Bring Down on the same row.
Expected: badge gone, tooltip back to normal, internet still works (the wlan owns the route again).

*** Screenshot View Image option
What we're verifying: the new post-capture menu entry opens the shot and puts its path on the clipboard (dispatch is unit-tested; this is the live end-to-end).
- Take a screenshot the usual way (region or fullscreen).
- Pick "View Image" from the fuzzel menu.
- Paste anywhere (the clipboard should hold the file path).
Expected: the shot opens in feh (the default image/png viewer) and the pasted text is the saved file's path under ~/pictures/screenshots/.

*** Wlogout square cells on velox (and later ratio)
What we're verifying: the exit menu's six buttons read as squares with visible muted borders, and only pointer hover shows gold (measured 361x361 px, but your eyeball is the arbiter).
- Press Super+Shift+Q.
- Look at the six cells' shape and borders; hover a couple of buttons; note the lock button is not gold before you hover it.
- Press Esc to close (careful: the letter keybinds are live — e is logout).
Expected: square cells with subtle borders, gold only under the pointer, muted ring on the focused button. Repeat on ratio after its sync.

*** wtimer: color states + click/scroll interactions on the live bar
What we're verifying: the timer module's interactions and CSS state colors render right on the live bar. The glyph, position (right of battery), countdown, and "+N" badge are already verified live; the per-state colors and the real mouse/scroll bindings are what's left. The logic is unit-tested (86 cases); this is the human-in-the-loop visual + input check.
- Left-click the timer module — a fuzzel menu offers timer / alarm / stopwatch / pomodoro; pick timer, enter =5s=.
- Watch it count down; at under a minute it should turn the urgent color (dupre orange #d47c59).
Expected: the timer reaches 0:00, a persistent notification fires, and the item disappears from the bar.
- Create two timers (e.g. =3m= and =10m=); a =+1= badge shows; scroll over the module.
Expected: scrolling cycles which item is primary (the displayed time/glyph changes); the badge count stays correct.
- Middle-click the module while a timer runs.
Expected: it pauses (dimmed paused color #5f5c52) and the countdown freezes; middle-click again resumes from where it left off.
- Right-click the module with items present.
Expected: a fuzzel menu lists the items; choosing one cancels it.
- Start a pomodoro (left-click → pomodoro); let a work phase elapse (or set short test phases by editing state).
Expected: the glyph + color switch between work (gold #d7af5f) and break (#8a9a5b), a notification fires at each phase change, and the cycle count advances.
*** Sysmon right-click cycles the visible metric (live waybar)
What we're verifying: right-clicking the collapsed sysmon module rotates the visible metric and the bar refreshes at once, left-click still opens btop, and the cpu/temp/mem icons render as real glyphs (not tofu boxes). The cycle logic is unit-tested; this is the live-waybar + visual confirmation.
- Reload waybar so it picks up the new =signal= / =on-click-right= config (Super+B relaunches it, or =pkill waybar; waybar &= from a terminal)
- Right-click the sysmon module several times, watching the visible metric
- Left-click the sysmon module once
Expected: each right-click advances the visible metric battery → cpu → temp → mem → disk → back to battery (velox is a laptop, so battery is in the ring) and the bar updates immediately. Every metric shows a sensible icon plus its value, no tofu. Left-click still opens the btop popup. The tooltip still lists all metrics.
*** Give the README a final read before public release
What we're verifying: =README.md= reads cleanly and accurately for a first-time reader, with no stale personal info and consistent public-fork placeholders.
- Open =~/code/archsetup/README.md=
- Read it end to end as if you've never seen the project
Expected: every section is accurate, the personal-project disclaimer reads right, the placeholders (=<your-domain>=, =github.com/yourusername=) are consistent, and nothing personal leaked into the public-facing draft.
*** Bt console: connect / disconnect a paired device
What we're verifying: a paired row's primary action toggles the real connection and the row's lamp + battery dial follow (the smoke drives the widgets against fakes; this is a real device).
- Open the bt panel (left-click the bar's bluetooth module, or Super+Shift+B).
- On a paired-but-disconnected audio device, click its row's connect action.
Expected: the device connects, the row lamp goes live, and if it reports battery a dial fills in with its level.
- Click the same row again to disconnect.
Expected: the device disconnects, the lamp dims, and its battery dial returns to NO DEVICE.
*** Bt console: rename a paired device
What we're verifying: the ✎ affordance renames a real device through the bluez Alias and the new name persists (mechanism live-probed on the M650 during the build; this is the in-panel path).
- In the bt panel, click the ✎ on a paired device's row.
- Enter a new name in the dialog and confirm.
Expected: the row's name updates to the new alias immediately, and it survives closing and reopening the panel (verify-after read confirms it stuck).
- Rename it back to its original name the same way.
*** Bt console: pair a nearby device (passkey flow)
What we're verifying: pairing a new device from the NEARBY list runs the pair flow into the passkey-confirm dialog and default-deny holds (this can't be auto-driven — it needs a real discoverable device and the passkey compare).
- Put a bluetooth device into pairing mode.
- In the bt panel, press SCAN and wait for the device to appear under NEARBY.
- Click its row to start pairing.
Expected: a passkey-confirm dialog appears; confirming completes the pairing and the device moves to PAIRED; dismissing it leaves the device unpaired.
*** Bt console: arm-forget a paired device
What we're verifying: the ✕ two-click arm-to-fire removes a real pairing (the arm latch is unit-tested; this confirms the real forget lands).
- In the bt panel, click the ✕ on a device you can re-pair later.
Expected: the row arms (tinted, a confirm affordance) rather than forgetting on the first click.
- Click again to confirm.
Expected: the device is unpaired and drops off the PAIRED list.
*** Bt console: discoverable toggle and adapter power switch
What we're verifying: the discoverable chip and the faceplate power switch drive the real adapter state.
- In the bt panel, click the discoverable chip.
Expected: the chip reads "discoverable on" and another device can see this adapter while it's on; clicking again turns it off.
- Flip the faceplate adapter-power switch off, then on.
Expected: off powers the adapter down (faceplate word → OFF, rows empty), on brings it back (POWERED, paired rows return). Airplane mode overrides — if it's on, the switch refuses with the way out.
*** Bt console: LOW BATT badge with a real sub-15% device
What we're verifying: a connected device reporting under 15% battery drives the faceplate LOW BATT badge and the red dial (the threshold is unit-tested; this needs a real device actually low).
- Connect an audio device whose battery is genuinely under 15% (drain one, or catch it low).
Expected: its battery dial renders red, and the faceplate shows the LOW BATT badge. Charge it above 15% and the badge clears on the next refresh.
*** 2026-06-28 Sun @ 12:54:47 -0400 Live-update guard verified on velox (live Hyprland)
Verified the =hypr-live-update-guard= PreTransaction hook end-to-end on velox
with Hyprland running (pid 1997). velox predated the feature, so the guard was
absent — placed =/usr/local/bin/hypr-live-update-guard= (755) and
=/etc/pacman.d/hooks/hypr-live-update-guard.hook= (644), byte-matching the
archsetup hyprland-step install. The guard now ships on velox permanently.

Results:
- Quick contract (=printf 'mesa\nhyprland\n' | guard=) → exit=1, BLOCKED banner,
  sorted pkgs, correct TTY remedy + sentinel path.
- Not-running branch (=HYPR_GUARD_RUNNING=0=) → exit=0, silent.
- Env override (=HYPR_ALLOW_LIVE_UPDATE=1=) → exit=0.
- Sentinel (=touch /run/archsetup-allow-live-gpu-update=) → exit=0; removed →
  re-armed exit=1.
- Real firing through pacman: =sudo pacman -S mesa= (same-version reinstall =
  Upgrade op on a guarded target). pacman ran the hook, fed =mesa= via
  =NeedsTargets=, the guard aborted, =AbortOnFail= stopped the transaction
  ("no packages were upgraded"); mesa unchanged at 1:26.1.3-2. This is the
  authoritative proof pacman parses + wires the hook.
- Full-logout end-to-end (guard quiet, upgrade completes after logout): covered
  by construction — the not-running branch exits 0, and a 0-exit PreTransaction
  hook lets pacman proceed normally (proven by the mesa abort showing the hook
  path runs). Not re-run under a real logout; no separate residual.
*** Wallpaper survives relogin (waypaper --restore)
What we're verifying: the hyprland =exec-once= now runs =waypaper --restore= instead of a hardcoded =awww img=, so a wallpaper chosen via =set-wallpaper= / waypaper / dirvish persists across a relogin. The exec-once only fires at Hyprland startup, so this can't be confirmed without a real relogin. (Mechanism already verified: =waypaper --restore= applied the persisted wallpaper via the awww backend, exit 0.)
- Set a wallpaper different from the current one (or pick one in waypaper, Super+Shift+P):
#+begin_src sh :results output
set-wallpaper ~/pictures/wallpaper/trondheim-norway.jpg
#+end_src
- Log out of Hyprland and back in (or reboot)
Expected: the wallpaper you just set is what comes back after login — not whatever was showing before, and never the old hardcoded default unless that's what you set.

*** velox per-host env applies after Hyprland restart
What we're verifying: the velox tier's env lines (GDK_SCALE/QT_SCALE_FACTOR 1.5, XCURSOR_SIZE 36) only apply at Hyprland startup, and the foot font moved to host.ini — neither can be confirmed over ssh.
- On velox, log out of Hyprland and back in (or reboot)
- Open a foot window — text should render at 12pt (same as before the migration)
- Launch Zoom (ideally from a browser link) — it should open at normal size with no per-app patch
- Check the cursor isn't tiny on the HiDPI panel
Expected: foot at 12pt, Zoom normally sized, cursor 36px — all from the velox tier, no local real files involved.

*** Dupre Chrome theme renders correctly
What we're verifying: the new Chrome theme's colors look right in a real browser — palette mapping can't be eyeballed from a manifest.
- Open chrome://extensions in Chrome
- Enable "Developer mode" (top right)
- Click "Load unpacked" and select =~/code/archsetup/assets/color-themes/dupre/chrome-theme/=
- Look at the window frame, toolbar, tab strip, and a new tab page
Expected: near-black frame (#151311), dark toolbar/omnibox (#252321), gold links on the new-tab page, steel-gray inactive tab text — coherent with the rest of the dupre desktop.

*** 2026-06-10 Wed @ 17:46:34 -0500 velox post-trim reboot verified; realtek firmware restored
Craig rebooted velox (passphrase at console); checks ran over SSH after boot. Wifi connected, TLP active, graphics fine. One dmesg hit: r8152 failed to load rtl_nic/rtl8156b-2.fw — the Framework Ethernet expansion card (RTL8156B) is Realtek, so the trim list wrongly dropped linux-firmware-realtek (a Realtek laptop camera is on USB too). Reinstalled the package on velox (its hook rebuilt the initramfs) and removed realtek from archsetup's trim list. The driver worked even without the blob (internal-defaults fallback), so this was correctness, not breakage.

*** Super+F dirvish popup: launch, float, dismiss-on-focus-loss, q
What we're verifying: the physical keychord opens a floating Dirvish popup; opening any file launches it independently (never inside the popup frame) and the popup auto-dismisses when focus leaves; navigating dirs keeps it; q is the manual close. The Wayland focus event that drives the auto-dismiss can't be driven headlessly — only a real keypress + real app launch confirms it.
- Press Super+F
- Expected: a Dirvish frame opens floating and centered, rooted at ~/ (home)
- Navigate into a directory with RET (or right-arrow)
- Expected: the popup stays open and shows that directory (browsing keeps it up)
- Open a video with RET (or o)
- Expected: the video opens in its player and the popup vanishes on its own — no q needed, nothing left in the way, and q never lands on the video
- Press Super+F again, open a PDF or image with RET
- Expected: it opens in zathura / feh (externally), NOT inside the popup frame; popup dismisses
- Press Super+F again, open a .txt or .org file with RET
- Expected: it opens in a NEW emacsclient frame (separate from your working session), not adopted into your current session; popup dismisses
- Press Super+F, then click another window without opening anything
- Expected: the popup dismisses on focus loss
- Press Super+F, then press q
- Expected: the popup closes completely (manual dismiss still works; no empty leftover frame)
- Press Super+F, then press Super+F again while it's still open
- Expected: still exactly one popup — the second press focuses the existing one, no second frame, no stray buffer (for several independent file managers, use C-x d)
- Press Super+Shift+F
- Expected: GUI nautilus opens (the binding nautilus moved to)
*** Network module Phase 1 — indicator states on the live bar
What we're verifying: =custom/net= shows the right state for each real network condition. The engine logic is unit-tested; this is the live-bar + visual check (states can't be faked on the running bar). Phase 2-3 tests get added under this task as those phases land.
- Reload waybar to pick up =custom/net=. Super+B does NOT reload a running bar — it only toggles visibility (SIGUSR1), and the bar reads a generated runtime config, so a stale copy keeps the old module. The correct reload regenerates the runtime config then restarts:
#+begin_src sh :results output
waybar-active-config && killall waybar && waybar-toggle
#+end_src
- On a normal connected network, read the module.
- Expected: wifi glyph + signal + SSID; tooltip shows IPv4, gateway, throughput, and a recent "online" probe result.
- Join the hotel/captive network (or any portal network).
- Expected: the module shows the captive state (distinct glyph + warning color), tooltip names the portal host.
- Unplug to a network with no internet (or block egress).
- Expected: the no-internet state (distinct from captive and from disconnected).
*** Network module Phase 1 — net doctor recovers rfkill from a TTY
What we're verifying: the console-recovery path works with no GUI, and recovers the framework's post-power-loss soft-block.
#+begin_src sh :results output
rfkill block wifi        # simulate the soft-block
rfkill list wifi         # confirm Soft blocked: yes
#+end_src
- Switch to a TTY (Ctrl+Alt+F3) and log in (no Hyprland).
#+begin_src sh :results output
make -C ~/.dotfiles online   # or: net doctor --fix
#+end_src
- Expected: doctor reports the rfkill block, runs =rfkill unblock wifi= + =nmcli radio wifi on=, reconnects, and ends "online" — all from the TTY.
*** Network module — bar clicks + airplane keybind (FINAL scheme)
What we're verifying: the custom/net clicks and the airplane keybind. Clicks (settled with Craig over live use 2026-06-29): left = =net-panel= toggle (the GTK panel), middle = =net portal= (floating terminal), right = =net-fix= (notify the doctor result when one-way; open a terminal only when fixable). Airplane = Super+Shift+A.
- Left-click =custom/net=.
- Expected: the GTK connection panel toggles open (left-click again, or Esc, closes it).
- Right-click =custom/net= while online.
- Expected: a desktop notification "Network / Online" (success), no terminal. When a repair is needed it instead opens a terminal running =net doctor --fix=. (Craig confirmed the notification delivers, 2026-06-29.)
- Middle-click =custom/net= on a captive network.
- Expected: =net portal= runs in the floating terminal — reset + opens the portal page.
- Press Super+Shift+A.
- Expected: airplane engages (wifi off, dim, low-power); =custom/net= shows the airplane glyph in gold. Super+Shift+A again restores everything.
- Check =airplane-mode= is still present (=ls ~/.local/bin/airplane-mode=), and =waybar-airplane= / =waybar-netspeed= / =custom/airplane= are gone.
*** Network module Phase 3 — panel Diagnose / Repair / Speed test tabs
What we're verifying: the four-tab panel works end to end. Left-click =custom/net= to open it.
NOTE (2026-07-04 audit): the "four-tab panel" framing predates the instrument-console rebuild (dotfiles 800ef60, f4e688e dropped the Blueprint pages). Diagnose/Repair/Speed-test now live in the single-screen console, not tabs — reword the steps to the console layout before running. Don't delete; the underlying behaviors still need verification.
- Diagnose tab → "Run diagnose".
- Expected: a list of steps (link, DHCP, gateway, DNS config, DNS resolution, internet) each with a ✓/✗/… glyph + evidence; on a captive network an "Open portal" button appears.
- Repair tab → click Reset (or Bounce, or DNS override test).
- Expected: a confirmation dialog with the exact wording (Reset names the network + new-MAC warning; Bounce "links drop briefly"; DNS test "reverts automatically"). Proceed opens a floating terminal that runs the repair (sudo prompt there) and shows the step output incl. cleanup-verified for the DNS test.
- Speed test tab → "Run speed test" (uses ~30s + data — do it on real wifi, not the metered hotspot).
- Expected: ↓/↑ Mbps + ping + server shown inline.
- Byte-rate→Mbps unit: VERIFIED 2026-06-30 (velox). Raw =speedtest-go --json= dl_speed read ~3.66M, unambiguously bytes/s (29 down / 80 up Mbps); =net speedtest= reported 33.62 / 77.99 through the wired path. =BYTES_PER_SEC = True= + =* 8 / 1e6= are correct, no flip needed. Remaining here is only that the panel renders the inline result.

** DOING [#B] Prepare for GitHub open-source release
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Remove personal info, credentials, and code quality issues before publishing.
*** 2026-07-09 Thu @ 16:32:54 -0500 Audit reconcile: docs/ scrubbed; scripts and history still open
The =docs/= half of "remove personal information" landed (=a8f7e9b=). Four absolute =/home/cjennings= paths became the repo root, a home-relative path, or a named variable — every snippet stays runnable. Three =~/projects/home= references and three =.ai/sessions/= links, all dead in any clone but Craig's, were replaced by what they mean. No =file:= links broke.

The scrub also turned up something the epic didn't name: nine tracked files carried =#+AUTHOR: Craig Jennings & Claude=, a co-author who is not a person and which survives conversion into docx, a wiki page, or a PDF. The five under =docs/= are fixed. *Four remain, and they are Craig's call* (dated records under =assets/=, which the rule says stay as history, but which are tracked and would publish): =assets/2026-06-19-collapsible-waybar-sides-spec.org=, =assets/2026-07-03-instrument-console-panels-build-summary.org=, =assets/outbox/2025-11-08-keyring-fix-next-steps.org=, =assets/outbox/2025-11-08-test-failure-analysis.org=.

The two open children below are untouched: the =scripts/= personal-information pass, and the git-history secret scrub.
*** 2026-06-16 Tue @ 00:55:39 -0500 Six dotfiles-scoped sub-tasks moved to the ~/.dotfiles project
Per the 2026-06-16 task audit, the six sub-tasks targeting files now owned by the standalone =~/.dotfiles= repo were handed off to that project (newly bootstrapped as its own AI project) and removed from this epic: "Remove credentials and secrets from dotfiles", "Remove/template personal info from dotfiles", "Remove binary font files from repo", "Move battery out of waybar sysmonitor group", "Resolution-adaptive scratchpad sizing", and "Dynamic waybar/foot config based on screen resolution". Handoff: =~/.dotfiles/inbox/2026-06-16-0053-from-archsetup-dotfiles-release-prep-handoff.org=. This epic now covers archsetup-proper release work only (scripts personal-info, device-specific config, history scrub, shellcheck, SPDX headers, README/LICENSE). The 2026-06-09 reconciliation note below is the prior state.
*** 2026-06-09 Tue @ 19:21:36 -0500 Reconciliation: six sub-tasks now target the ~/.dotfiles repo, not archsetup
Phase 3.2 removed the in-repo =dotfiles/= tree, so six sub-tasks below no longer describe archsetup content — they target files now owned by the =~/.dotfiles= repo (=git.cjennings.net/dotfiles.git=): "Remove credentials and secrets from dotfiles", "Remove/template personal info from dotfiles", "Remove binary font files from repo", "Move battery out of waybar sysmonitor group", "Resolution-adaptive scratchpad sizing", and "Dynamic waybar/foot config based on screen resolution". Their paths are relative to that repo now. Kept here for tracking per Craig (2026-06-09); he'll re-scope the archsetup-vs-dotfiles split shortly. archsetup-proper release work (scripts personal-info, device-specific config, shellcheck, and scrubbing the pre-=b10cba5= dotfiles secrets from archsetup's own history) stays this task.
*** 2026-05-11 Mon @ 13:01:29 -0500 AI Response: Open-source-prep source audit
Checked each subtask below against the source / git state. Bottom line: almost nothing is fully done. =LICENSE= and =README.md= were added this session (see those subtasks); the rest still stands.
- *Remove credentials and secrets from dotfiles* — NOT DONE. All five named files still tracked: =dotfiles/common/.config/.tidal-dl.token.json=, =.config/calibre/smtp.py.json=, =.config/transmission/settings.json=, =.msmtprc=, =.mbsyncrc=. =.gitignore= lists none of them; no =.example= templates exist.
- *Remove/template personal info from scripts* — PARTIALLY DONE. Repo URLs ARE config-driven (=archsetup:141-146= use =${dwm_repo:-https://git.cjennings.net/...}=, documented in =archsetup.conf.example=). Still personal: =archsetup:2-3= (email/website header), =init:8,21= (=root:welcome=), =scripts/post-install.sh:17-56= (personal repos).
- *Remove/template personal info from dotfiles* — NOT DONE. =.gitconfig= has =c@cjennings.net=, =name = Craig Jennings=, =github user = cjennings=, =safe.directory= and employer creds; =.config/mpd/musicpd.conf= + =mpd.conf= still use =~cjennings/= / =/home/cjennings/= paths; =.ssh/config= has personal/employer hosts; =.config/yt-dlp/config:2= has =c@cjennings.net=; =hyprland.conf:3= has personal attribution.
- *Scrub git history of secrets* — NOT DONE. 275 commits; history not fresh, no filter-repo evidence.
- *Remove device-specific configuration* — NOT DONE. =archsetup:1486-1493= still creates the Logitech BRIO udev rule unconditionally; no config flag.
- *Add README.md for GitHub* — DONE (this session — initial draft, pending review). See subtask below.
- *Add LICENSE file* — DONE (this session — GPL-3). See subtask below.
- *Remove binary font files from repo* — NOT DONE. =dotfiles/common/.local/share/fonts/= still tracks 8 PragmataPro =.ttf= files, =AppleColorEmoji.ttf=, and other commercial fonts (Cartograph, MonoLisa, ComicCode, etc.).
- *Make claude-code installation optional* — NOT DONE. =archsetup:1817-1818= runs =curl -fsSL https://claude.ai/install.sh | sh= unconditionally; no flag.
- *Add input validation for username and paths* — PARTIALLY DONE. =archsetup:326-328= validates =$username= against =^[a-z][a-z0-9_]*$= (plus reserved-names check, marked DONE separately). No validation of =$source_dir= or other path vars.
- *Move battery out of waybar sysmonitor group* — NOT DONE. =dotfiles/hyprland/.config/waybar/config:27-37= still has =battery= inside =group/sysmonitor=.
- *Resolution-adaptive scratchpad sizing* — NOT DONE. No size/move windowrules for scratchpads in =hypr/conf.d=.
- *Dynamic waybar/foot config based on screen resolution* — NOT DONE. No resolution-detection/generation script.
- *Bulk shellcheck cleanup* — PARTIALLY DONE. =shellcheck archsetup= still shows 68 findings: 30×SC2329, 16×SC2174, 15×SC2024, 4×SC2086, 1 each SC2155/SC2129/SC2005. The 4 SC2086 (unquoted) are the ones a reviewer would flag — those are the priority.
- *Document testing process in README* — NOT DONE. =scripts/testing/README.org= exists but isn't the project README. (Now unblocked — root README exists.)
- *Add guard for rm -rf on constructed paths* — DONE 2026-05-20. All three constructed-path deletes routed through a =safe_rm_rf= guard (absolute / no-=..= / inside-allowed-prefix / real-dir checks); unit-tested in =tests/safe-rm-rf/=.
- *Standardize boolean comparison style* — NOT DONE. Mixed: =[ "$var" = "true" ]= at =archsetup:542,544,569= vs bare =if $var;= form ~7 places elsewhere.
- *Replace eval with safer alternatives* — NOT DONE. =archsetup:442= still =if eval "$cmd" >> "$logfile" 2>&1;= in =retry_install=.

*** 2026-06-28 Sun @ 13:34:03 -0400 Cancelled: calendar-feed URL rotation
Craig's call — not rotating. The three private iCal URLs (Google personal, Proton with PassphraseKey, Google DeepSat) sat in git history from =500b1f5= (2026-05-13) until the 2026-05-20 filter-repo scrub, which removed them from local + remote history. The residual exposure is only to anyone who cloned the repo in that 2026-05-13..05-20 window; Craig accepts that window rather than regenerating all three tokens on ratio. The history scrub already happened; the live =calendar-sync.local.el= is owned by the emacs project. Closing without rotation.

*** 2026-05-20 Wed @ 12:09:32 -0500 Scrubbed the calendar secret from git history
=dotfiles/common/.emacs.d/calendar-sync.local.el= (private Google/Proton/DeepSat ical URLs, added in =500b1f5= for stow distribution) was discovered while folding tmux-util into stow. Sent the file back to the emacs project's inbox, =git rm='d it, then =git filter-repo --invert-paths --path= purged it from all 29 affected commits. Force-pushed (=0921e4d...618e6cc=, with lease) and ran =reflog expire= + =gc --prune=now= on the bare repo at =/var/git/archsetup.git=. Verified: the file is in zero commits, the secret tokens return zero matches across all history, and =500b1f5= / =0921e4d= are unreachable on both local and remote. Rotation of the URLs tracked as the sibling TODO above. This also proves =filter-repo= works cleanly here — relevant precedent for the broader [[*Scrub git history of secrets (or start fresh)][history-scrub task]] below (the 5 credential files are still in history).

*** TODO [#B] Remove/template personal information from scripts
- =archsetup= lines 3-4: personal email and website in header
- =scripts/post-install.sh=: personal git repos and server URLs (the old =scripts/gitrepos.sh= was consolidated into this script in =dae7659=, so its personal =git.cjennings.net= clone targets now live here)
- =init= line 9: hardcoded password =welcome=
**** 2026-06-28 Sun @ 13:29:29 -0400 Reconciled: dotfiles repo URLs already config-driven
Dropped the "lines 141-146 hardcoded =git.cjennings.net= URLs" bullet. archsetup:138-140 reads =DOTFILES_REPO= / =DOTFILES_BRANCH= / =DOTFILES_DIR= overrides (defaults only, documented in =archsetup.conf.example=), so that item is already done. Refreshed the stale line numbers on the remaining bullets (header email/site now lines 3-4, init password now line 9, after the SPDX headers shifted the files).

*** TODO [#B] Scrub git history of secrets (or start fresh)
Even after removing files, secrets remain in git history.
Options: =git filter-repo= to rewrite history, or start a fresh repo for the GitHub remote.
Recommend: fresh repo for GitHub (keep cjennings.net remote with full history).
**** 2026-06-28 Sun @ 13:29:29 -0400 Reconciled: 589 commits, 5 credential files still in history
History is now 589 commits (the 2026-05-11 note's "275" is stale). Only the calendar-feed file has been filter-repo'd so far (2026-05-20). The five credential files remain in history at their pre-=b10cba5= paths: =.tidal-dl.token.json= (5 commits), =calibre/smtp.py.json= (6), =transmission/settings.json= (5), =.msmtprc= (8), =.mbsyncrc= (9). None are tracked in the current tree. The scrub-or-fresh-repo decision still stands.
***** 2026-07-04 Sat @ 11:48:24 -0500 Count refresh — history now 565 commits; re-verify the 5-file claim before scrubbing
The 2026-07-04 audit found the history is now 565 commits, down from the 589 recorded above. Because the count dropped, re-verify that the five credential files are still present in history (re-run the per-file =git log --all -- <path>= check) before relying on the scrub scope — the earlier count is stale and the file set may have moved.

*** 2026-06-24 Wed @ 19:41:56 -0400 Gated device-specific udev rules behind a flag
The Logitech BRIO udev rule is now wrapped in =if [ "$install_device_udev_rules" = "true" ]=, fed by a new =INSTALL_DEVICE_UDEV_RULES= key (default yes, opt-out — still mainly a personal project). Added the var default, the config read, a =validate_config= check, and an =archsetup.conf.example= entry. Verified: default/yes writes the rule, no skips it, bogus is rejected; =bash -n= clean.

*** 2026-06-28 Sun @ 13:37:33 -0400 Added README.md — full draft complete, final read filed
=README.md= is substantively done at repo root (10.9 KB), covering project description, features, requirements, installation, the =archsetup.conf= configuration guide, security considerations, contributing, and license, with generic placeholders for the eventual public fork. The 2026-05-11 "first pass" note below is superseded. Craig's final read before public release is filed under "Manual testing and validation"; closing as code-complete pending that human check, per the audit rule.

**** 2026-05-11 Mon @ 13:01:29 -0500 AI Response: Initial README draft
Drafted =README.md= at repo root, modeled on =~/code/chime/README.org=. First pass — review and run a voice/style pass before committing. Personal info (emails, =cjennings.net= URLs, personal repo names) intentionally replaced with placeholders for the eventual public release.

*** 2026-05-19 Tue @ 01:54:29 -0500 Added GPL-3 LICENSE file at repo root
GPL-3 chosen. Canonical GPLv3 text landed at =LICENSE= on 2026-05-11 (commit =f80e664=). README already links to it. SPDX/license headers across source files (or a NOTICE file) split out as a new sub-task below for the eventual public release.

*** 2026-06-24 Wed @ 19:41:56 -0400 Added SPDX headers to all shell scripts
Swept =# SPDX-License-Identifier: GPL-3.0-or-later= in right after the shebang of all 24 shell scripts in the repo (=archsetup=, =init=, =scripts/**/*.sh= incl. =scripts/testing/=). The dotfiles are a separate repo now, so they aren't swept here. Verified the header sits at line 2 (after the shebang) and syntax still passes.

*** 2026-06-09 Tue @ 19:21:36 -0500 Made claude-code install optional
Shipped in =f2dad22= (feat: make the claude-code install optional). The =curl | sh= from claude.ai now sits behind a config flag instead of running unconditionally.

*** 2026-06-09 Tue @ 19:21:36 -0500 Input validation added (validate_config + validate_username)
validate_config + validate_username shipped (detail in the 2026-05-11 note below). The =$source_dir= path check was judged unnecessary — it derives from the now-always-validated =$username= (=/home/$username/.local/src=). Closed as done.

**** 2026-05-11 Mon @ 18:20:49 -0500 AI Response: validate_config + validate_username added
Added two pre-flight validators to =archsetup= (right after =load_config=, before any install step):
- =validate_username()= — the lowercase / starts-with-letter / =[a-z0-9_]= / not-reserved check, extracted from the inline block in =preflight_checks()=. Fixes an existing gap: the inline check only ran on the *prompted* path, so a config with =USERNAME=root= (or =USERNAME=foo bar=) slipped through unvalidated. Now both =preflight_checks= and =validate_config= call it.
- =validate_config()= — runs whenever =--config-file= is used: rejects unknown =DESKTOP_ENV= (must be dwm/hyprland/none) early instead of dying in step 7-9; rejects =AUTOLOGIN=/=NO_GPU_DRIVERS= values that aren't =yes=/=no= (currently silently ignored); basic shape check on =LOCALE=; and a scheme + no-whitespace/no-leading-dash check on the six =*_REPO= URLs that get passed to =git clone= (rejects e.g. =--upload-pack=…= injection). Plain =echo …>&2; exit 1= (the logging helpers aren't defined that early). =$source_dir= needs no separate check — it's =/home/$username/.local/src=, derived from the now-always-validated =$username=.
Not a security boundary (=load_config= sources the config as bash; a hostile config can already run anything) — it's typo-catching. Verified with =bash -n= and a smoke-test matrix of good/bad inputs through both functions. The next =make test= run confirms valid configs still install. Leaving as DOING for review.

*** 2026-05-20 Wed @ 06:50:25 -0500 Swept shellcheck across the shell scripts
Census across the 16 shell scripts (=archsetup=, =init=, =scripts/*.sh=, =scripts/testing/=): 124 findings, zero errors. Triaged against "what matters for public review" and confirmed the 2026-01-24 read — most are intentional or documented-acceptable:
- SC2024 (14, sudo redirects), SC2174 (16, =mkdir -p -m=), SC1091 (13, unfollowable sources), SC2329 (32, functions invoked indirectly via the =STEPS= dispatch array), SC2153 (1, =DISK_PATH= sourced from =vm-utils.sh=) — all false positives or accepted.
- SC2086 on =$SSH_OPTS= in =vm-utils.sh= (×4) and =$TEMP_DISKS= in =cleanup-tests.sh= — intentional word-splitting; quoting would break them. The SSH_OPTS-as-array refactor is the proper fix, deliberately deferred (codebase-wide, one atomic change).
- SC2086 integer tests in =[ ]= (=archsetup=, =cleanup-tests=) — safe, note-level style; left to avoid churn in the just-fixed =retry_install=.
- SC2015 (×2, =vm_exec && success || warn=) — =success=/=warn= return 0, so C won't spuriously fire. Idiomatic.

Fixed the four that are genuine: =init= (a =#!/bin/sh= script) used =$(</etc/hostname)= (SC3034 bashism → =$(cat ...)=) and an unquoted =$interface_up= (SC2086 → quoted); =shellcheck init= now clean, =sh -n= passes. Suppressed the two =VM_IP= SC2034 warnings with documented =# shellcheck disable= directives (consumed by the sourced =validation.sh=, which shellcheck can't follow). 124 → 120; the remaining 120 are the triaged-acceptable set above.

*** 2026-05-20 Wed @ 06:32:17 -0500 Documented the testing process in the README
The README only covered the VM integration harness; the unit-test layer under =tests/= (Python =unittest=, fake-binary-on-PATH, one dir per script — =layout-navigate=, =tmux-util=) was undocumented. Added a =make test-unit= target that runs every =tests/*/test_*.py= suite explicitly (=unittest discover= can't find them — hyphenated dir names aren't valid package paths), then rewrote the README Testing section into "Unit tests" and "Integration tests (VM harness)" subsections, including how to add a suite for a new script. Updated Contributing to point at =make test-unit= for script changes. 61 unit tests pass via the new target.
*** 2026-05-20 Wed @ 18:22:42 -0400 Added safe_rm_rf guard on constructed-path deletes
Added a self-contained =safe_rm_rf <path> <allowed_prefix>= helper to =archsetup= and routed all three constructed-path deletes through it. The guard refuses to run unless the target is absolute, free of =..=, deeper than a bare top-level dir, strictly inside the allowed prefix (not the prefix itself), and a real directory (not a symlink); otherwise it prints the reason and returns non-zero without deleting. On the happy path it delegates to =rm -rf=.

Sites converted (the line numbers in the original task body were stale — actual sites located by grep):
- =--fresh= state-dir wipe — prefix =/var/lib/archsetup=.
- =git_install= clone-retry cleanup (=build_dir= under =$source_dir=).
- =aur_installer= yay clone-retry cleanup (same prefix).

The helper is defined before the top-level =--fresh= handler (which runs at load time, before the logging helpers exist), so it carries no =error_warn= dependency and reports refusals to stderr itself. The two in-function sites keep their existing =|| error_warn= / =|| error_fatal= handling.

Tests: =tests/safe-rm-rf/test_safe_rm_rf.py= sources the real function out of the script and exercises Normal/Boundary/Error cases (13 tests) against real temp dirs. =make test-unit= green (61 tests), =bash -n= clean, no new shellcheck warnings.
*** 2026-06-24 Wed @ 19:41:56 -0400 Standardized boolean comparisons on the explicit form
Rewrote the bare =if $var= boolean conditionals (=show_status_only=, =fresh_install=, =skip_gpu_drivers=, =detected_intel/amd/nvidia=, plus two =! $var= negation chains) to the explicit =[ "$var" = "true" ]= / =!= "true"= form, and quoted the one unquoted =install_claude_code = true=. Left =if $step_func= alone — that's the STEPS function-dispatch, not a boolean. Verified: only =step_func= remains bare, all comparisons are quoted, =bash -n= clean.

*** 2026-05-26 Tue @ 15:27:09 -0500 eval task moot — the line-434 eval is gone, the survivor is deliberate
Verified: the only =eval= left in =archsetup= is line 578 in =retry_install=, and it's intentional and documented — it captures =$?= directly from =eval "$cmd"= to dodge the if-compound-swallows-exit-code trap. Replacing it with an array would reintroduce that bug. The line-434 eval this task pointed at no longer exists. Nothing to change.

** TODO [#B] The audio doctor never checks the microphone :bug:audio:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-13
:END:
The classifier is output-only. =diag.probe_semantic= already collects =default_source= and =default_source_present=, and =classify.py= reads neither: the word "source" appears once in the whole module, in the graph row that counts them. So a muted mic, a default source naming an unplugged device, or a mic at zero volume all classify as =healthy=, and the verdict prints "the default output is present and audible" while the input side goes unexamined. Found 2026-07-10 while asking whether the doctor would have caught Chrome losing the mic. It would not have.

Not a scope decision. The spec's Non-Goals never say the doctor is output-only, and its Summary calls it a doctor for "a broken sound stack".

Work: mirror the sink rules onto the source. =probe_semantic= gains =default_source_muted= / =default_source_volume=; the classifier gains the source cases; =classify.findings()= gains an input row beside its =defaults= row, so the CLI wall and the GUI wall both grow it for free.

Two things not to get wrong. An absent microphone is legitimate on a desktop, so "no input devices" must never be a fault the way =no-output-devices= is. And a monitor source is a legitimate default source (recording desktop audio), which is why =probe_semantic= passes =include_monitors=True= — inheriting the panel's display filter here would call a working setup broken.

Specced 2026-07-10 after discussion with Craig, and the design grew past the original gap: [[file:docs/specs/2026-07-10-audio-doctor-input-side-spec.org][docs/specs/2026-07-10-audio-doctor-input-side-spec.org]] (DRAFT, four decisions open). A doctor key per direction, a kernel-level capture probe below PipeWire, PTT-aware muting, and a direction-aware guard. The precedence question the build would have faced is gone: a doctor per direction means the user's press says which side they came to fix.

Parent spec: [[file:docs/specs/2026-07-09-audio-doctor-spec.org][docs/specs/2026-07-09-audio-doctor-spec.org]] (IMPLEMENTED). This is a v1 gap found after the fact, not a phase of it.

** TODO [#D] Telega coredump recurrence tell :bug:maint:
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:

*** 2026-07-09 Thu @ 16:32:54 -0500 Audit reconcile: the fix is holding; re-graded C → D
Verified rather than assumed: =~/.telega/telega-server.log= carries zero =tdat_plist_value= assertions, and the newest telega-server coredump is 2026-07-08 16:57 — before .emacs.d upgraded the elisp. Nothing has recurred.

Re-graded =[#C]= → =[#D]= per the bug matrix. There is no defect to fix here; it is a watch item with a named tell, and the severity × frequency read is cosmetic (host coredump noise on a metric we own) × rare edge case → P4 → =[#D]=. It stays on the list only so the tell isn't lost.
The maintenance console's coredump metric flagged telega-server on ratio (8 coredumps) and velox (18). Root cause was a version skew: the Dockerized =zevlg/telega-server:latest= is frozen at the 2026-06-05 build while the installed elisp lagged at 20260513, so the newer server's plist parser choked on the older elisp's output. .emacs.d fixed it by upgrading telega to 20260706 on both machines (docker kept, =docker pull= is a no-op against the frozen image). Host-coredump pollution should stop. If zevlg later pushes a =:latest= that outruns the installed elisp, the skew and the coredumps recur — the tell is a fresh =tdat_plist_value:500= assertion in =~/.telega/telega-server.log=. The durable escape is a host-native pinned TDLib build, at the cost of an AUR source build.

* Archsetup Resolved

** DONE [#B] Maintenance console build :feature:
CLOSED: [2026-07-08 Wed]
:PROPERTIES:
:SPEC_ID: 9d9df833-c592-4aec-a7df-50d588e943ce
:END:
Build the maintenance console per [[file:docs/specs/2026-07-07-maintenance-console-spec.org][maintenance-console-spec.org]] (DOING; review incorporated 2026-07-07 — two rounds, all 13 decisions DONE, all 10 findings DONE, verdict Ready; Craig approved config paths + sysmon right-click re-homing). Design source of truth: [[file:docs/design/maintenance-console-design-ideas.org][maintenance-console-design-ideas.org]]; pixel reference: [[file:docs/prototypes/2026-07-07-maint-console-E5-selector-subpanel.html][E5 prototype]]. CLI-first =maint= package in dotfiles (archsetup-owns-dotfiles applies: edit, test, commit, push, note dotfiles inbox per phase); thresholds TOML + VM harness in archsetup. Each phase gates on green tests before its commit.

*** 2026-07-07 Tue @ 20:11:23 -0500 Built Phase 1 — package skeleton, thresholds, contract (dotfiles 43a39ac)
maint/ package + shim + tests/maint/ (48 tests, fake-tool harness); thresholds two-layer merge with disable flags + env overrides; capability probe (battery ⇔ type==Battery, verified against ratio's Mains+USB tree); =maint status --json= with disk-usage/pacman-cache/failed-units pilots. Seeded configs/maintenance-thresholds.toml in archsetup from the workflow's values and installed a copy to ~/.config/archsetup/ on ratio (installer wiring stays Phase 13). Full make test green (50 suites). Live on ratio: worst=warn from the real 10.75 GB cache; du grades on its printed total, not its exit code (root-owned partial-download dirs make it exit 1) — pinned in a test.

*** 2026-07-07 Tue @ 20:49:05 -0500 Built Phase 2 — storage & snapshot collectors (dotfiles 94e8371)
All Phase 2 collectors TDD'd against the fake-tool harness (argv-matched cases added): btrfs unallocated/scrub-age/device-errors (SMART cross-check), per-disk SMART health + last self-test, fstrim posture (discard mount option = continuous trim, ratio's real setup), disk top-consumers on a new slow-local cache (=maint scan --slow=, ~/.local/state/maint), snapper counts split timeline·single·pre-post (split keys on cleanup, not type), full ZFS family. =priv.py= pulled forward from Phase 6 (read-only verbs =smart_json= + =btrfs_scrub_status=, validated args, MAINT_SUDO) so probes stay elevation-free. maint suites 48→118; make test green (51). Live read-only verified on BOTH hosts: ratio worst=warn (10.75 GB cache; scrub 0d, SMART ok), velox over tailscale shows the ZFS family live (ONLINE/autotrim-on/scrub 1d/231 snaps) with a genuine fstrim warn (timer off — Phase 6 Confirm remedy). TOML gained storage keys (btrfs_unalloc_warn_gb, smart_*, fstrim_stale_days, hog_*); installed copies refreshed on ratio + velox.

*** 2026-07-07 Tue @ 21:30:59 -0500 Built Phase 3 — packages, security, systemd collectors (dotfiles 3eaab6e)
All Phase 3 collectors TDD'd: orphans name+size with [curation.kept_orphans], pacnew safe-delete vs needs-merge, keyring freshness, reboot-required (uname -r vs modules dir), -Qkk on =maint scan --slow= with a noise split (unverifiable-as-user reads, mtime-only, [curation.qkk_known] paths tallied but never counted — an unfiltered live run sat permanently red at 68 "altered" files; classified, ratio shows 10 real findings); =maint scan --net= writes checkupdates/yay/arch-audit/fwupd caches (yay rc-1-silent = AUR-clean zero; failures never clobber good cache; malformed payloads degrade to unprobed), readers grade with age + stale bump; is-system-running with state-aware cause, failed-unit roster since/exit/journalctl hint, maintenance-timers meta-metric (capability-derived expected set, @-prefix instance match), taint letters decoded. Prereq done: arch-audit added to archsetup deps + installed both hosts. Review subagent (live-verified) found 2 Important (yay AUR-clean rc quirk, cache-shape envelope kill) + 2 minor — all fixed red-first; archsetup now enables btrfs-scrub@-.timer on btrfs installs (was ratio-only out-of-band, the meta-metric expects it). maint suites 118→169; full make test green. Live on ratio: 23-metric envelope, worst=warn from real findings (10.75GB cache, 8 orphans, mirrorlist.pacnew, 713 repo updates, 32 CVEs, 1 firmware).

*** 2026-07-07 Tue @ 21:59:27 -0500 Built Phase 4 — logs + mem·pwr collectors (dotfiles 40c448b)
All Phase 4 collectors TDD'd: journal error digest (identifier groups w/ first/last seen + journalctl hint, known-noise curation tallied but never counted, byte-array MESSAGE decoded), coredumps by binary over a TOML window, kernel/hw event classes (MCE/EDAC/IO/thermal/GPU — generic driver noise excluded), journald disk usage, app-log staleness mirroring the log-cleanup cron's filename-date rules; memory pressure + top-5 RSS, OOM kills (kill-verdict lines only — one event spans several matching lines), swap/zram, temps via sensors -j (k10temp Tctl / coretemp package / amdgpu edge), Intel thermal-throttle count, EPP read, battery capacity/health/charge-cap, unclean-shutdown rate from wtmp (last -x pairing). Throttle/EPP/battery self-gate on sysfs and drop out of the envelope. Review subagent (live-verified) found 1 Blocking — windowed-empty coredumpctl exits 1 (my earlier --since-now pin was wrong), which left the metric permanently unprobed on healthy hosts — plus OOM ~3× overcount and a malformed-curation-entry envelope crash; all fixed red-first, re-review Approve. maint suites 169→227; full make test green. Live read-only both hosts: ratio 34 metrics (telega-server crashing 3× in window — real find), velox 38 (genuine 75% battery-health warn; charge_control_end_threshold ABSENT on velox's cros_ec hardware → Phase 9's SET 80% lever must gate on the file). TOML +journal_disk_warn_gb/coredump_window_days, new [memory], [power] temp/health/unclean keys; installed both hosts.

*** 2026-07-07 Tue @ 22:35:28 -0500 Built Phase 5 — network posture + services collectors (dotfiles 2e92a9f)
All Phase 5 collectors TDD'd after live-pinning every tool format on both hosts (velox presence sweep over tailscale: full tool set present): nm_state (passive — NM's own connectivity check covers DNS+HTTP, so status never generates traffic), chronyc offset vs [network].ntp_offset_warn_ms, tailscale Health + peer tally, fail2ban per-jail bans, listeners digest via priv ss (expected-entry curation tallied-not-counted, unexpected+wildcard=crit, v4/v6 dedup, multi-holder sockets match any name), firewall (priv ufw status; fallback requires unit-active AND ufw.conf ENABLED=yes; names "ufw down — N public binds exposed" from the digest), rsyncshot freshness (1MB log-tail parse; hourly graded only where the log shows hourly runs — velox is daily-only; daily ≥48h/missing = crit), docker df JSONL + stopped containers w/ [curation.expected_containers], libvirt roster (queried only when libvirtd already active — a read must not socket-activate), cron drift (root rsyncshot expectations gate on backup-source). priv.py +4 read-only verbs; thresholds.curation_entries() tolerant helper. Review subagent (live-verified): Request Changes — 1 Important (malformed fail2ban counts raise ValueError into the envelope) + ufw-disable unit-state false positive, docker unknown-unit sizes reading 0, unfloatable chrony offset; all fixed red-first, re-review Approve. maint suites 227→297; full make test green (54 suites, unpiped exit 0 — an earlier tail-piped run masked make's exit code, rerun properly). Live both hosts: ratio 45 metrics, velox 49. REAL FINDS: listeners crit on both (ratio postgres/8080 wildcard binds, velox syncthing *:8384/*:22000) — curation material for Craig; backups stale on both since 2026-07-06 (truenas down for repairs, Craig confirmed — clears when it returns). TOML +[network]/hourly-backup/cron keys, installed both hosts.

*** 2026-07-07 Tue @ 23:14:56 -0500 Built Phase 6 — remedy layer, priv verbs, guard, doctor CLI (dotfiles e171c08)
The full remedy layer TDD'd in one pass (no 6a/6b split needed): remedies.py allowlisted table (~30 remedies — exact argv, tiers per the design tables as amended by the dated decisions, re_probe_ids, long flags, RECLAIM SPACE macro resolving steps against the envelope incl. per-config snapper expansion); priv.py +28 write verbs (closed builders, type-validated args, per-verb timeouts; =build()= returns the exact sudo-prefixed argv so dry-run and the GUI arm-press share one string); guard.py pure live-update matcher over new TOML =[updates] guard_patterns= (fnmatch, case-sensitive, pairwise-tested; cold pending-cache is surfaced as a note, never silent); curation.py user-layer writes (mark/unmark/clear + KEEP/MARK KNOWN/EXPECTED wrappers, disable-flags for shipped defaults, self-verifying TOML emitter); doctor.py clean|review + iter_fix streaming the event feed (begin/ok/fail/guard/note; failures red without aborting; re-probed truth after each action); cli =maint fix <id> [--dry-run] [--force]= + =maint doctor clean|review= wall renderer; status envelope now stamps =levers= (always-on for free controls: epp_set, charge_limit, update, topgrade). KILL four guards: pid:name victim, TERM-only verb, session-critical denylist ([curation.session_critical], new TOML table) at preview, comm revalidation + denylist re-check at fire. Live-pinned before coding: coredumpctl HAS NO clean verb (find -mtime +N -delete via priv), log-cleanup at ~/.local/bin/cron/, rsyncshot takes <mode> <keep>, ufw enable needs --force, EPP value enum. Review subagent (live-verified): Request Changes — 1 Blocking (json.dumps astral-char surrogate pairs = invalid TOML: one emoji in a MARK KNOWN example would brick thresholds.load() for the whole console → ensure_ascii=False + write self-verifies by re-parsing before replace) + 1 Important (KeyError from sparse installed TOML escapes the event stream → caught, degrades to a red wall entry) + 2 Minor (cold-cache guard note, comm-vs-denylist re-check); all fixed red-first (6 new tests), re-review Approve. maint suites 297→393 (96 in the new suite); full make test green ×2 (unpiped, exit 0). VERIFIED LIVE on ratio: doctor clean executed the Auto tier for real — paccache -ruk0 freed 223.91 MiB and the re-probe measured the cache drop exactly; the live-update guard trips on ratio's genuinely-pending mesa/vulkan-radeon set. TOML +guard_patterns/+session_critical/+timeline_quarterly installed both hosts.

*** 2026-07-07 Tue @ 23:52:40 -0500 Built Phase 7 — GUI shell (dotfiles d28435e)
The GTK4 window lands as the fourth panel sibling (net.cjennings.maintpanel, humble-object layering: GTK-free PanelModel presenter + pure viewmodel formatters + thin gui.py view). Read-only shell: faceplate (worst lamp, CVE/ATTN badges, MNT·01, close), inert doctor/update keys, updates strip (pending/CVE/AUR/firmware from the strip ids), eight-band selector (lamp + fixable/watch split from the envelope's levers; updates/security categories fold into PACKAGES), metric-row subpanel per band with per-id value formatting (bytes/percent/days/°C/ms/bool). Hydration off-thread; 3s live tier (build_live: cheap mem·pwr reads) + 30s full rebuild, both visibility-gated; MAINT_PANEL_FIXTURE=good|bad renders shipped envelope fixtures (reshaped from the E5 prototype by tests/maint/gen_fixtures.py, deterministic, real remedy ids) and freezes both tiers so the degraded board stays put on a healthy machine. Review subagent: Request Changes — 1 Blocking (build_live returned bare metrics, so the 3s merge stripped the levers the full envelope established and the band splits oscillated between tiers → attach_levers in build_live) + 2 Minor (synthetic error envelope missing worst read as healthy green → panel.error_envelope with worst=unprobed; smoke leaked the panel process on mid-run assertion errors → try/finally reap); all fixed red-first, re-review Approve (+1 hardening: merge into an error envelope is a no-op). maint suites 393→447 (54 in test_panel.py); full make test green (56 suites, unpiped exit 0). AT-SPI smoke PASS on live Hyprland (faceplate, badges, strip, all 8 tiles, band switch, clean close; window parked on a special workspace, never the active one). Screenshots off-workspace via headless output: bad fixture + live board vs the E5 reference. Live board on ratio honest: 32 CVE / 13 ATTN, NETWORK red from the real listeners crit, SERVICES fixable from the truenas-down backup staleness. GOTCHAS: windowrules need hyprctl reload before first launch (panel opens tiled otherwise); pkill -f "maint panel" self-matches the invoking shell — anchor with "maint panel$".

*** 2026-07-08 Wed @ 00:39:23 -0500 Built Phase 8 — GUI levers: digests, armed keys, rotary band, live strip (dotfiles 388521e)
The storage/snapshots/packages subpanels grew their evidence digests (disk hogs, per-drive SMART, device errors, snapper by-type, orphans KEEP/REMOVE, pacnew DELETE/MERGE, CVE advisories), packages got the E5 rotary band selector (ORPHANS/PACNEW/ADVISORIES), and the updates strip went live: state-tiered border, UPDATE/TOPGRADE behind the arm-to-override live-update guard, REBOOT offered when required or after an update lands. Every lever is arm-then-fire — first press shows the exact argv on the new act line (identical to =maint fix --dry-run=, the parity criterion), second press fires through =doctor.iter_fix= with the re-probe landing back on the act line; one key armed at a time, band/deck switch disarms; fixture boards always dry-run and never write curation, which is what lets the AT-SPI smoke exercise the full press-press path safely. Engine support: snapper probe emits =stale_singles= (DELETE STALE's candidates, newest kept), both scrub probes parse running-%, =doctor.pending_updates()= public for the arm-press guard read, fixtures regenerated with deterministic evidence. Review subagent (live-verified): Request Changes — 2 Important (refresh tiers raced a firing remedy: a 30s tick could probe the package DB mid-transaction and land stale data after the fire's own re-probe → tiers pause on =firing= + post-fire rebuild retries past an in-flight tick; digest builders KeyError'd on identity-less evidence rows replayed from cache files → skip-and-degrade, 8 new red-first tests) + 4 Minor (all taken: broadened strip-key except, scrub key offers first idle target instead of vanishing while any runs, busy-refused press no longer consumes the arm, fire-time guard refusal names the tripped set in GUI wording). Re-review: Approve. maint suites 447→516; full =make test= green (exit 0, unpiped); AT-SPI smoke PASS end-to-end on the bad fixture; headless-output screenshots vs E5; live board honest on ratio (real orphans w/ KEEP/REMOVE, 10.5 GB cache CLEAN, arm-press names the genuinely pending mesa set). Velox pulled to 388521e (no restow/TOML/windowrule changes). Scroll-position-on-arm check added to Manual testing.

*** 2026-07-08 Wed @ 01:51:26 -0500 Built Phase 9 — GUI: systemd, logs, mem·pwr, network, services (dotfiles 9a2985f + a3da942)
Shipped in two gated halves. 9a (9a2985f): failed-unit + timer rosters (armed RESTART/RESET per unit; ENABLE only on present-but-inactive timers), logs rotary deck (SIGNAL / KNOWN NOISE / COREDUMPS / KERNEL·HW) with the journal MARK KNOWN lifecycle (arm-then-fire mark bound to identifier + sample, UNMARK, armed CLEAR MARKS, empty-sample withhold), OPEN JOURNAL terminal delegation, row levers VACUUM/CLEAR/RUN. 9b (a3da942): CPU-mode segmented control (current + "default" excluded — priv's closed verb refuses "default"), top-memory guarded KILL (session-critical disabled pre-press, arm refusal behind it, fire-time comm revalidation behind both), battery SET 80% gated on the sysfs knob (velox: unsupported → withheld, verified live), listeners curation (probe emits pid + owning unit from the cgroup v2 line; STOP for system.slice services, KILL otherwise, "?" watch-only), firewall ENABLE / tailscale UP / chrony STEP row levers, services deck (containers MARK EXPECTED + START, docker df, cron roster + backups w/ worst-of chip lamp). RECLAIM SPACE deferred to Phase 10's wall by design — the act line can't honestly render a 7-step stream. Reviews: 9a Request Changes (curation done-callbacks bypassed the busy-retry rebuild → routed through _refresh_after_fire) → Approve; 9b Request Changes (DEFAULT EPP dead control + cgroup hybrid-hierarchy, "?"-curation, cron chip lamp, unprobed ✓ minors — all fixed red-first) → Approve. maint 516→612 tests; make test 58 suites exit 0; AT-SPI smoke PASS across all 8 bands; headless screenshots vs E5; live envelopes verified both hosts.
**** 2026-07-08 Wed @ 01:13:48 -0500 Built 9a — systemd + logs subpanels (dotfiles 9a2985f)
Failed-unit roster (armed RESTART/RESET per unit, since/exit context, journalctl hint) + expected-timer roster (ENABLE only on present-but-inactive timers); logs rotary deck (SIGNAL / KNOWN NOISE / COREDUMPS / KERNEL·HW) with the journal MARK KNOWN lifecycle (arm-then-fire mark bound to identifier + sample, UNMARK, armed CLEAR MARKS, empty-sample withhold), OPEN JOURNAL terminal delegation, and row levers VACUUM / CLEAR / RUN. Review subagent (live-verified): Request Changes — 1 Important (curation done-callbacks called refresh_full, which silently no-ops while a tick refresh holds busy → a mid-tick write left a pre-write envelope on the board up to 30s while the act line claimed success; routed all four incl. Phase 8's keep through _refresh_after_fire) + minors taken (single _mark_token definition, three test gaps closed, docstring truthfulness, stale arm wording cleared on band/deck switch — my own find during screenshot eyeballing). Re-review: Approve. maint 516→562 tests; make test 58 suites exit 0 unpiped; AT-SPI smoke PASS incl. the new rosters/deck/mark path; headless screenshots vs E5. velox pulled (no restow). Accepted gap: KNOWN NOISE subsec lists the live curation table on a fixture board — fixture-sourced entries list is later polish.

*** 2026-07-08 Wed @ 02:36:54 -0500 Built Phase 10 — results wall + doctor keys (dotfiles cc2fb5d)
Results wall shipped as the session action log: doctor events stream per-event (begin opens a running entry with the exact argv, ok/fail resolves in place, re-probe notes follow), 3.5-visible-row scroll cap with autoscroll, HIDE/COPY (COPY emits CLI-wall text), curation writes logged (fixture-suppressed as dim notes). CLEAN UP streams the auto tier with no arm press; REVIEW & FIX swaps the subpanel to the confirm-tier roster (item-less remedies get armed FIX keys, per-item ones point at their subpanel; review events carry item + worst severity). RECLAIM SPACE macro landed on the disk-usage row (deferred from 9b); macro arm line shows first step + remaining count; zero-step macros refuse by name. Review (live-verified): Request Changes — 1 Important: the roster's FIX for update/topgrade could never override the live-update guard (fire-time re-arm used the bare rid while the FIX key arms on "rid:", and _on_lever never passed force — an infinite arm→guard loop); fixed by threading the pressed token through _fire and sharing the strip's guard arming (_guard_arm_line + panel.guarded). Minors all taken: wall_abort resolves stranded running entries red when a stream dies, empty-macro IndexError → named RemedyError, test predicate tightened. Re-review: Approve, no residuals. maint 612→645; make test 59 suites exit 0 ×2; AT-SPI smoke extended (wall, HIDE/SHOW, CLEAN UP stream, review roster, 14-line RECLAIM stream) PASS; screenshots vs E5; LIVE verify on ratio: real CLEAN UP on the live board streamed "3 done" with real re-probes (coredumps warn 5, disk_usage ok 60). Velox pulled cc2fb5d.

*** 2026-07-08 Wed @ 03:24:44 -0500 Built Phase 11 — waybar glyph + timers (dotfiles 10033be)
custom/maint replaced custom/sysmon: waybar-maint renders from the state file maint-scan.timer writes every 30 min; color tracks the worst diagnostic (lever-less) metric only, actionable findings never color the bar; missing/stale scan data dims to an honest unprobed state naming the timer. Battery hosts show live battery % + charging icon from sysfs on waybar's interval; low charge forces red. Left-click toggles the console (new maint-panel wrapper); right-click keeps the btop scratchpad. Bare =maint scan= became the glyph scan (envelope → state → RTMIN+12); maint-net-scan.timer drives =--net --slow= hourly. Retired same commit: waybar-sysmon, sysmon-cycle, both suites, #custom-sysmon CSS ×3 — plus unnamed collateral the grep caught: waybar-collapse's collapsed base set (maint glyph stays on battery hosts — it IS the battery display now) and a stale dotfiles todo. Review (live-verified): Approve with 4 Minor, all taken red-first — scope!=Device gate so a wireless mouse's battery can't masquerade as the machine's (shared capability.device_scope), unprobed diagnostics named in the tooltip instead of "all clear" under a dimmed glyph, comment truth, dead state field dropped; re-confirm clean (677 maint tests). Full make test exit 0; smoke PASS (first run's pipe masked a transient AT-SPI race — rerun unpiped clean). LIVE both hosts: ratio timer fired for real (glyph amber from genuine diagnostics: firmware update + journal errors; 11 actionable NOT coloring), velox pulled + one-time stow + battery 100% text w/ warning class. velox's user systemd manager found wedged (pre-existing) — timers enabled via manual symlinks + state seeded; Craig rebooting to clear.

*** 2026-07-08 Wed @ 04:15:45 -0500 Built Phase 11b — prototype fidelity pass (dotfiles 3ee22a8)
Rebuilt the GUI presentation to the E5 instrument-card idiom. Each band renders a 4-column card grid: big-number readouts (viewmodel.card_spec registry — one builder per bespoke metric id, count-card fallback for the rest), cairo progress bars w/ threshold ticks (cache scaled to 1.5× the warn line), radial gauges (scrub cadence, keyring/topgrade age, temperatures — sweep scaled to the metric's own threshold), status chips (PASSED/FAILING/INACTIVE/YES; firewall chips the state word, exposure detail moves to the caption), and the CPU-mode segmented control promoted from digest section to card (current mode lit, alternatives as joined armed keys — same DIGEST_ONLY epp_set arm-then-fire). Evidence digests stay full-width below the grid. Selector became a two-row 4-col grid of wide tiles w/ count chips (3! · 7✓, crit reddens) + severity left borders; subpanel header carries the attention/ok/fixable/watch summary right-aligned beside a gold band title. The scoped NVMe wear dial was skipped honestly — no standalone wear metric exists (wear rides SMART evidence rows). Presentation only: engine/remedies/doctor/curation and the lever contracts unchanged. New 42-test suite (card kinds, threshold scales, running-scrub override, unprobed degrade incl. tick drop, seg exclusions, chips/subhead/band_counts, both-fixture sweep). Review (live-verified): Approve w/ 1 Important (pre-existing: chrony "not running" string crashed format_value's ms branch — fixed w/ card guard) + 4 Minor all taken red-first (disk None guard, _pct negative clamp, unprobed tick, dead row_detail/split_label/.maint-val removed); re-review Approve. make test 61 suites exit 0 ×2 unpiped; AT-SPI smoke 92 checks PASS (new gauge/chip/subhead assertions — DrawingAreas carry accessible_role=IMG + names so the smoke can see them); all 8 bands + review roster screenshotted on both fixtures + the live board against the settled headless-Chrome prototype render. Velox pulled (no restow). Gotcha pinned: a card body's vexpand propagates up a Gtk.Grid — a one-card band ballooned until grid.set_vexpand(False).

*** 2026-07-08 Wed @ 05:05:20 -0500 Ran the granted /refactor sweep over the maint package (dotfiles a178470..73e9d94)
Four parallel read-only scans (complexity, duplication, dead-code, simplification) converged on the same hotspots; ten behavior-preserving commits applied and pushed. Headlines: six curation handlers folded into one _curation helper (~150 lines of repeated fixture/done plumbing, act/wall strings byte-identical), _on_lever/_on_strip_key merged into _press_lever, _on_fired split (_rearm_after_guard + _fire_summary), the "tool: not found / exit N" wording single-sited in cmd.why() across 20 probe sites, exact clones deduped (packages curation_entries, _parse_asctime, double VERSION), ten uniform digest sections folded into _evidence_digest, _card_body split per kind, listener key rule named, plus a small-simplifications batch. Deliberately skipped: format_value's flat chain (readable as-is), _journal_lines generalization (rc-semantics risk), test-panel-maint Makefile target (Phase 12 scope). Verification: suite parity exact pre/post (59 suites / 2490 tests, exit 0 unpiped), AT-SPI smoke PASS end-to-end, live ratio envelope 45 metrics zero unprobed, review subagent walked the full diff against 3ee22a8 and approved (no behavior change; accessible names untouched; its one latent-robustness minor pinned as an invariant comment). gui.py 1680→1605; net −103 lines. Velox pulled.

*** 2026-07-08 Wed @ 05:46:02 -0500 Built Phase 12 — VM remedy scenarios + AT-SPI target (archsetup d6993d3, dotfiles 0636554)
Scenario orchestration landed as =run-maint-scenarios.sh= over the existing vm-utils.sh snapshot primitives: nine break → =maint fix= → assert scenarios in three groups (logs, packages, systemd), batched per Decision 11 (one boot per group, maint-ready snapshot restored between groups, clean-install restored at the end). =run-maint-nspawn.sh= is the fast lane — the packages group against a cached pacstrap rootfs in seconds. Plan layer is pure (=--list=, no KVM) with a 19-test unit suite; =make test-maint= wires it in. Dotfiles side: panel smoke now runs GOOD + BAD fixture passes (healthy render, badge tallies from the fixture, no-REBOOT, leak check) and =make test-panel-maint= is enumerated. Verified: full VM run 9/9 green from pristine snapshot ×2 (test-results/maint-20260708-054029), nspawn 4/4, smoke 100 checks exit 0, unit suite green under both FS_PROFILE values. Review subagent: Request Changes (FS_PROFILE env leak in the unit suite, journal-vacuum assert vacuous both directions) — all six findings fixed red-first, re-review Approve. FS_PROFILE=zfs scenarios: filtered but unexercised — the zfs base image fails to build (see the zfs DKMS task).

*** 2026-07-08 Wed @ 06:18:58 -0500 Built Phase 13 — install wiring, workflow move, docs (archsetup bef7053 + 18c081f, dotfiles 9a3f0c7)
Installer: install_maintenance_config in user_customizations installs the shipped TOML to ~/.config/archsetup/ (re-run refreshes it, ~/.config/maint/ user layer never touched) and enables maint-scan/maint-net-scan timers via wants-symlinks (hyprland-only — the units live in that stow tier; no session bus during install, syncthing idiom); orchestrator + dispatch-branch tests pin the wiring. Deps swept from the probes' argv: expac, lm_sensors, fwupd added (arch-audit was already in from Phase 3; the rest ride base/archinstall or existing sections). maint/README.md in dotfiles covers user (CLI, panel, glyph, config layers, guard, privilege) + developer (module map, env seams, four test layers, add-a-metric recipe). system-health-check.org moved home → archsetup docs/workflows/ and rewired: TOML-authoritative severity section, threshold lines cite TOML keys (TOML wins on disagreement), inventory paths → docs/homelab-inventory/, maint-CLI fast-path note; only the four #+HOSTNAME: host inventories moved (+ the truenas specs asset they link) — the personal gear records stay in home per the domain split; discoverability via .ai/project-workflows symlink + notes.org entry; handoff note in home's inbox (2026-07-08-0607) lists home-side removals and flags strix-soak-watch as a candidate to move later. Verified: fresh-shell =maint status= green from installed paths on ratio; make test-unit exit 0; dotfiles make test exit 0; shellcheck count unchanged; review subagent Approve (2 Minor doc nits, both taken).

*** 2026-07-08 Wed @ 06:18:58 -0500 Flipped the spec to IMPLEMENTED
Spec keyword DOING → IMPLEMENTED with a dated history line naming the closing commits (phases 1-13, dotfiles 43a39ac..9a3f0c7, archsetup d6993d3/bef7053/18c081f) and the Status mirror updated. Residuals tracked as their own tasks: manual-testing checklist, zfs base-image DKMS bug [#C], vNext [#D].
** DONE [#B] Signal and known-noise counts disagree with the journal :bug:maint:solo:
CLOSED: [2026-07-09 Thu]
Not a counting bug — a semantics bug (dotfiles eded0c0). journal_errors grouped by identifier and summed lines, but displayed one sample message, and logs.py overwrote that sample on every iteration so it named the *last* error rather than the counted one. Live on ratio: systemd had 10 error lines this boot — 6 "Failed to start Emacs text editor." and 4 org-roam auto-sync — which is exactly the 10-vs-6. Now grouped by identifier + normalized message (pids, hex addresses, uuids, long integer runs blanked), so a row's count is that message's count, and each row's hint carries a =--grep= reproducing it (verified: 6 and 4). MARK KNOWN binds the volatile-free literal instead of the pid-bearing sample, so marking suppresses exactly what the row counted. The KNOWN NOISE headline counts matched lines while the deck lists patterns; the caption now names both units.
Logs > SIGNAL reports 10 journal errors and the signal item claims 10, but the group holds one item and the open journal shows it 6 times. KNOWN NOISE reports 19 against 12 visible entries. The counts are the point of the band, so a wrong number undermines the whole diagnostic. Find where the tally diverges from the rendered set — likely counting matched journal lines rather than distinct grouped entries, and counting curated entries against a different window than the one displayed.
** DONE [#B] Persist the CPU pill setting across reboot :feature:maint:
CLOSED: [2026-07-09 Thu]
The kernel resets energy_performance_preference to the driver default every boot. A successful epp_set now records the preference in the maint user config; maint-epp-restore.service replays it at login. No-op when nothing is remembered, the host lacks EPP, or the mode already matches. A hand-edited bogus value is refused by priv's closed verb. Installer enables the unit via the existing wants-symlink idiom. Verified live on ratio: unit enabled, ran clean, EPP untouched. (dotfiles a7f34bd, archsetup 430ef1a)

The power/perf/balance pill sets EPP live, but the value reverts to its original setting after a reboot. Add a way to make a selection permanent. Needs a call on where persistence lives — a systemd unit, a udev rule, or a config file the maint package owns.
** DONE [#B] Preview the pacman + AUR update queue before upgrade :feature:maint:solo:
CLOSED: [2026-07-09 Thu]
A QUEUE key on the updates strip streams the pending pacman + AUR set onto the results wall, each package with its version move; `maint queue` prints the same in a terminal. The cap reports how many it dropped rather than truncating silently, and a cold cache says it has no pending set instead of rendering "nothing pending". Verified live against ratio's 105 pending packages. (dotfiles 8837a47)

Show which pacman and AUR packages are queued before UPDATE or topgrade runs, so the armed live-update guard isn't the first place the package set becomes visible.
** DONE [#B] Audio PTT state desync across panel, waybar icon, and actual state :bug:audio:solo:
CLOSED: [2026-07-09 Thu]
Two state stores and two copies of the logic. The panel flipped an in-memory ptt_armed flag with its own enter/exit rules; the CLI toggle and the bar tag read a runtime state file. The file is now the only state and ptt.toggle_plan the only decision. The panel adopts the persisted state on every refresh, routes its toggle through the shared plan, and persists before signalling waybar (writing after would race the bar into re-rendering the pre-toggle class). Tests drive both entry points against one state across armed/disarmed x muted/live. (dotfiles 9524a44)

The audio panel's push-to-talk, the waybar PTT icon, and the real PTT state disagree. All three must agree at all times. Cover every state transition with tests, not just the happy path — this is the second PTT bug after the pre-talk mute restore landed 2026-07-05 (dotfiles 1443b9e).
** DONE [#B] Publish the homelab inventory into the agent knowledge base :chore:docs:
CLOSED: [2026-07-09 Thu]
The four host inventories, the TrueNAS specs asset, and the ratio USB/xHCI record now live as roam nodes in =~/org/roam/hardware/=, linked from the "Homelab Hardware Inventory" index node. =docs/homelab-inventory/= is gone; roam is canonical.

Carried rather than pointed, per home's argument, which beat mine. Pointing keeps one canonical copy but breaks the write path: a third project that discovers a durable hardware fact would have to write into archsetup's repo to record it, which the cross-project rules forbid, so the fact would sit in an inbox instead of on the device's page. Moving answers the canonical-copy worry without that cost.

The =* Automated Capabilities= drawer moved with the pages. =system-health-check.org= now resolves a host's node by its =#+HOSTNAME:= keyword under =${ROAM_DIR:-$HOME/org/roam}/hardware/= — never by filename, since the timestamp prefix isn't stable. Verified: all four hosts resolve, an unknown host degrades to "NO INVENTORY FILE", and ratio's capability drawer still parses. A host without the roam clone (mybitch, truenas) takes the same no-inventory path it always did.

The strix kernel watch stayed at =docs/workflows/strix-soak-watch.org= — it's a workflow, not an inventory page.
** DONE [#C] Optional ticking sound for the pomodoro :feature:timer:
CLOSED: [2026-07-09 Thu]
WTIMER_TICK=1 plays a quiet tick each second through a pomodoro's work phase; off by default, silent through rest (a tick there would undo the break), while paused, and while awaiting a manual advance. Rides the bar's existing once-a-second `wtimer render` call. Playback is fire-and-forget so it can never stall or break the bar. Sound is an 18ms 1.4kHz sine at a tenth of full scale (notify/tick.ogg); WTIMER_TICK_SOUND overrides. Craig still has to judge whether the tick sounds right. (dotfiles 886dafd)

Timers module. Off by default; needs a sound choice and a volume decision.
** DONE [#C] Refresh indicator for the top-like items :feature:maint:
CLOSED: [2026-07-09 Thu]
A hairline under the top-memory section drains toward the next live refresh, so a slow board reads differently from a frozen one. Withheld on a fixture board, which never live-ticks — a countdown that never counts is worse than none. The fraction is a pure viewmodel function; only the cairo stroke lives in the GUI. (dotfiles 5d384d9)

Show when the process/memory rows are about to refresh, so a stale-looking board is distinguishable from a frozen one. Shape is a design call — a countdown, a pulse, or a progress hairline.
** DONE [#C] CPU pill should not reposition on selection :bug:maint:quick:solo:
CLOSED: [2026-07-09 Thu]
epp_rows dropped the current preference from the key list, so the remaining segments slid sideways on every selection. Every preference now renders in a fixed order with the current one lit and unpressable in place, at full contrast rather than GTK's disabled dimming. The separate lit label is gone — the active segment is the lit one. Three separate tests encoded the old contract (unit, AT-SPI smoke, phase11b card); all moved with the behavior. The smoke caught a real gap the unit tests missed: the EPP row renders through the card path, which ignored the disabled flag. (dotfiles 5d384d9)

The perf/balance/power pill moves position depending on which option is selected. The segmented control should hold a fixed layout and mark the active segment in place.
** DONE [#C] Session identifier needs more width :bug:waybar:quick:
CLOSED: [2026-07-09 Thu]
Not waybar — the tmux status line. status-left-length was never set, so tmux clipped the drawn status-left to its default of 10 columns: "[aiv-archsetup] " came out "[aiv-archs", losing the closing bracket and the separating space, which is why the window list ran into it. Set to 40. Applied to the live tmux server too. (dotfiles dc3c27f)

The session section truncates; give it more characters, and keep the window sections that follow from running into it. Screenshot: [[file:/home/cjennings/pictures/screenshots/2026-07-07_145236.png][2026-07-07_145236.png]]
** DONE [#C] minimal tier zsh login PATH gap :bug:dotfiles:quick:solo:
CLOSED: [2026-07-09 Thu]
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Fixed (dotfiles =c6a7878=). =minimal/.zprofile= added, mirroring =common/='s PATH-only prepend and its rationale comment.

Tested the invariant rather than the file, because the bug *was* a per-file fix missing a tier: any stow tier shipping a =.zshrc= must ship a =.zprofile= that prepends =~/.local/bin=. Red first, and it named =minimal=. Verified behaviorally as the task asked — stowed the tier into a throwaway HOME and read a real =zsh -lc 'echo $PATH'=, which now leads with =~/.local/bin=. Full =make test= green.

The 2026-07-08 fix gave common/ a .zprofile (zsh logins never read .profile, so ~/.local/bin was missing from TTY/ssh PATH — dotfiles 39cef40). The minimal/ tier has the same gap: it carries .zshrc and .profile but no .zprofile, so headless installs (DESKTOP_ENV=none stows minimal/ instead of common/) still get zsh logins without ~/.local/bin. Add a matching minimal/.zprofile (PATH-only prepend, same rationale comment). Verify: fresh minimal stow, zsh -lc 'echo $PATH' shows ~/.local/bin first. The installer itself is clean — it writes no shell profiles; the stow layer owns this.
** DONE [#B] Doctor button for the audio panel :feature:audio:
CLOSED: [2026-07-10 Fri]
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
What happens when pipewire or pulseaudio is broken — would we know? Spec: [[file:docs/specs/2026-07-09-audio-doctor-spec.org][docs/specs/2026-07-09-audio-doctor-spec.org]] — *READY*, all eight decisions closed 2026-07-09, Phase 0 shipped. Buildable: Phase 1 diag, Phase 2 classify, Phase 3 repair + doctor, Phase 4 the DOCTOR key and wall. Grounded in a live survey of ratio's stack: there is no PulseAudio (pipewire + wireplumber + pipewire-pulse, all user-scope, so no sudo anywhere), =pactl= hangs against a server that accepts and never answers, and the panel can't diagnose itself because =pactl= is the layer most likely to be down. Mirrors the net panel's diag/classify/repair/doctor split, not the maint console's priv-verb table.

Re-graded =[#C]= → =[#B]= by the 2026-07-09 audit: the spec is written, Phase 0 shipped, and six decisions are all that stand between this and a build. Active backlog, not parking lot.

*** 2026-07-09 Thu @ 14:12:00 -0500 Fixed: waybar's audio module died against a half-wedged server
My first read of this was wrong. =pactl.run()= already carries =timeout=8= and raises =PactlTimeout=, so the panel degrades and does not freeze. Correcting the record rather than leaving the claim standing.

The real bug, found by driving it: =build_status()= guarded its two device-list reads but called =default_sink()= and =default_source()= outside the guard. A server that answered the lists and then stopped answering raised =PactlTimeout= straight out of a function whose whole contract is to degrade to =ok: False=. =waybar-audio= calls it with nothing catching it, so the bar's audio module died instead of dimming; the GTK panel survived only because its background worker catches everything.

A server that hangs on the *first* call never exposes this — =list_sinks= raises inside the guard and masks the unguarded reads behind it — so the obvious regression test passes against broken code. =fake-pactl= grew =FAKE_PACTL_SLEEP_DEFAULTS= to express "answers the lists, hangs on the defaults". Verified end to end: =waybar-audio= now emits =class: degraded= / "Audio graph unavailable" and exits 0. dotfiles =4d42eb3=; full =make test= green.

*** 2026-07-10 Fri @ 06:41:00 -0500 Shipped: the audio doctor, phases 1 through 4
=audio diag=, =audio doctor=, =audio doctor --fix=, and the panel's DOCTOR section header with DIAGNOSE / FIX keys streaming into a results wall (copy + close). Spec is IMPLEMENTED. dotfiles =01a1d80=, =76857e3=, =7223c51=, =1550ac8=, =4320255=, =bd33440=.

Live verification on ratio replaced the planned VM run for remedies 1, 4 and 5, and found two bugs no fake could surface: =pactl get-default-sink= prints the literal =@DEFAULT_SINK@= placeholder when no default is set, and remedy 1 re-probed before WirePlumber had re-created the devices, calling a repaired stack broken. Both fixed.

Left open, not a v1 gap: mpv played silently while the doctor read healthy. The stack was genuinely fine, and per-application stream routing is an explicit spec Non-Goal. Worth a task only if it recurs.

Second sighting, 2026-07-10: Chrome stopped recognizing the microphone while the stack was healthy (Shure MV7+ default, unmuted, 82%, PTT off). Restarting Chrome fixed it; nothing on the machine was touched. Both sightings share a shape the doctor cannot see: the graph is fine and one client cannot use it. A third sighting turns this into a design question, namely whether the doctor should say "the stack is fine, the fault is in the application" rather than a bare healthy.
** DONE [#C] Net panel: Enterprise error never dismisses :bug:dotfiles:network:
CLOSED: [2026-07-12 Sun]
Fixed in dotfiles =a157bed=. Root cause: error toasts are sticky by design (so background refreshes can't wipe an unread error), but the enterprise join hint's flow posts no follow-up status and row clicks post none either, so nothing ever replaced it. Fix: a window-wide capture-phase click gesture dismisses a sticky toast on the user's next interaction; policy in =viewmodel.toast_action_plan= (unit-tested), timed toasts and background clears unchanged. Panel smoke run confirms launch/doctor/close with the gesture installed. Pointer-level dismiss is a manual-testing child (AT-SPI can't drive pointer gestures). Repro screenshot: =~/pictures/screenshots/2026-07-10_195911.png=.
** DONE [#C] Net diagnostics leak connection names + SSIDs into copyable report and --json :bug:dotfiles:network:solo:
CLOSED: [2026-07-12 Sun]
Resolved in dotfiles =df1543a=: the =redact_ssid= toggle now scrubs saved profile names, active SSIDs, envelope-carried names, and =.nmconnection= keyfile basenames from the copyable report and the diag/doctor =--json= envelopes (one systemic pass in =redact.py=; MAC/IP scrub applies to those envelopes too). On-screen output and functional envelopes (status/list) unchanged. 15 new tests; live-verified on ratio (toggle on removes the active connection name from =diagnose --json=, default unchanged).
The net doctor's copyable report (=report.py=, =scrub_text=) scrubs only MAC/IP, and =net diag/doctor --json= (=cli.py=) dumps the raw dict with no redaction. SSID redaction lives only in the event log (=redact_event=, gated on =redact_ssid=, default off). So a connection name (usually the SSID) appears in the clear in the link-step evidence and in every =--json= consumer — the copyable report is exactly the text a user pastes into a bug report. Secrets (PSK/password/token/portal URL) are already stripped, so this is names, not credentials: Minor severity, graded on severity alone per the privacy carve-out.

Split out of the 2026-07-11 net-doctor-expansion spec review: that spec's new rival-manager/keyfile-perms verdicts keep parity with this pre-existing behavior rather than half-solve it. Fix shape: extend redaction to cover the connection name + keyfile basename across the copyable report and =--json= (one systemic pass, not per-verdict), with a redaction test. Engine-wide, so it wants one coherent change rather than being bolted onto the expansion work.
** DONE [#B] Bt doctor expansion v1 — build the READY spec :feature:dotfiles:bluetooth:
CLOSED: [2026-07-12 Sun]
:PROPERTIES:
:SPEC_ID: 3d4d61c4-e5df-44e9-b8e0-40b31452c3f7
:END:
Build the [[file:docs/specs/2026-07-11-bt-doctor-expansion-spec.org][bt doctor expansion]] (IMPLEMENTED). Adds a dmesg firmware-hint probe (names the missing blob on a no-adapter fault) and a boot-enablement probe (catches an adapter disabled at boot) to the shipped bt doctor (=~/.dotfiles/bluetooth/=). Archsetup owns the dotfiles work end to end. All phases shipped and fake-verified (d19fdca, f05a9b4, d7d859f); the live reboot-persistence half is on the manual-testing checklist.
*** 2026-07-11 Sat @ 03:06:32 -0500 Built the two read-only probes
New module =bluetooth/src/bt/probes.py= plus =doctor.py= wiring, on dotfiles main (=d19fdca=, pushed). Two reads the diagnose chain never did: =firmware_hint()= scans the current boot's kernel log for per-vendor firmware-load failures (Intel ibt-*.sfi, MediaTek BT_RAM_CODE, Realtek rtl_bt, Broadcom .hcd, Qualcomm QCA), returning the named blob via a bounded =cmd.run(journalctl -k)= that reuses the =doctor.py:84= precedent; =boot_enablement()= reads three boot-persistence signals (bluez AutoEnable from main.conf [Policy], =systemctl is-enabled bluetooth=, whether TLP lists bluetooth in =DEVICES_TO_DISABLE_ON_STARTUP=). =diagnose()= gates the firmware read to the no-adapter branch and the boot read to the soft-blocked/powered-off branch, so a healthy run reads neither; the raw signals ride a new =probes= key that =doctor()= carries into =--json=. Detection only: no verdict, formatter, or repair change (that's Phase 1). Every read degrades to None on an unreadable tool/file, so a probe that can't see never invents a fault. AutoEnable absent/unset reads None, not false, matching bluez's compiled default of true, so only an explicit =AutoEnable=false= is the fault. New env roots for tests (=BT_MAIN_CONF=, =BT_TLP_CONF=, defaulting to absent temp paths in the Sandbox base so no test reads real /etc); =fake-journalctl= branches on =-k=, =fake-systemctl= answers =is-enabled bluetooth=. 125 bt tests, full =make test= green; =/review-code= approved (no Critical/Important; one Minor noting the firmware read also covers the btctl-unavailable branch, harmless). Inbox note sent to dotfiles.
*** 2026-07-11 Sat @ 03:16:59 -0500 Built the firmware-hint Guide verdict
On dotfiles main (=f05a9b4=, pushed). The no-adapter step now names the blob: a new =_no_adapter_step= consults =probes.firmware_hint()= on a genuine no-adapter fault and, on a per-vendor signature match, sets =evidence= to "no Bluetooth adapter found — <Vendor> firmware <blob> failed to load" and =next_action= to "update linux-firmware and reboot", tagged with a new =code="no-adapter-firmware"= so a =--json= consumer can branch without string-matching. A clean log keeps the generic hardware/driver verdict. A Guide, not a repair: the step carries no =repair= action, so =--fix= never touches it, and it needs no privilege model (so Phase 1 lands independently of the shared cross-panel model). A missing =bluetoothctl= (=BtctlError=) short-circuits before the firmware read, so the verdict fires only on a real no-adapter fault, not a broken install — this also tightened Phase 0 (which read the log on both None branches) to the genuine no-adapter case. =_mk= gained a uniform =code= key (default None) added to every diagnose step, mirroring the existing =repair= key; no test asserts an exact step key-set, verified. =format_doctor_human= already renders =evidence=/=next_action=, so no formatter change. 133 bt tests (+8), full =make test= green; =/review-code= clean. Inbox note sent to dotfiles.
*** 2026-07-11 Sat @ 06:48:21 -0500 Built the persistent-power verdict + fix
On dotfiles main (=d7d859f=, pushed). =_powered_step= consumes the bt Phase 0 boot-enablement probe: AutoEnable explicitly false, service disabled at boot, or TLP listing bluetooth → =powered-off-persistent= (code + evidence naming the cause) carrying the =persist-power= repair; absent/unset config reads as auto-enable-on (bluez default) → the plain =power-on=, so healthy machines can't false-positive. The =persist-power= repair fixes only the causes set (AutoEnable=true, =systemctl enable bluetooth=, drop bluetooth from the TLP list), powers the adapter on now, and verifies each cause cleared. Config edits are pure idempotent text transforms in a new =bootconf= module (comments preserved), staged and installed via a fixed-destination =cp= verb so the write needs root but the mutation is unit-testable. =priv.py= gained three narrow verbs (=enable-bluetooth=, =write-main-conf=, =write-tlp-conf=). The repair is Privileged and resolves through =panelkit= before running: can't-elevate degrades to the guide. The bt shim gained =panelkit= on its path. 287 bt tests + 65 make-test suites green, review-code Approve, voice. Inbox note sent. Live half (real reboot persistence) is the VM/manual checklist.
*** 2026-07-12 Sun @ 09:14:00 -0500 Flipped the bt spec to IMPLEMENTED and logged the vNext items
Spec status heading now IMPLEMENTED (dated history line + Status mirror); all four phase headings DONE. vNext items (stale-bond signature, connection-parameter hints, bt-audio-profile expansion) logged as the "Bt doctor vNext" task. The live half (real reboot persistence) remains on the manual-testing checklist — findings there come back as bugs.
** DONE [#B] One copy + close control pair on every output wall :feature:dotfiles:solo:
CLOSED: [2026-07-12 Sun]
Resolved in dotfiles =dccd744=: every wall carries the o-copy/o-clear overlay pair. bluetooth gained the copy key (transcript via the new =viewmodel.step_copy_line=, CLI-shaped); maint traded its header COPY key for the overlay pair, kept HIDE, and its ✕ clears the session log via =PanelModel.wall_clear=; the four hand-rolled wl-copy calls collapsed into =panelkit.clipboard.copy_text= (PANELKIT_WLCOPY test seam), moving maint off the GTK clipboard so its copies survive the panel closing too. 15 new tests (8 clipboard, 4 step_copy_line, 3 wall_clear); full suite 66 green; all four panel smokes run live off-workspace — maint + audio fully pass, net + bt fail only the pre-existing state-word startup race.

Converge all four instrument-console output walls on the net panel's well controls: a copy glyph and a ✕ close, as an overlay at the top right, hidden until content lands. Craig's call, 2026-07-10, while reviewing the audio doctor's wall: "network panels as standard across all others, make it consistent."

Where they stand today, no two alike. net has copy + ✕ (=net/src/net/gui.py=, the =o-copy= / =o-clear= overlay). bluetooth has ✕ but no copy. maint has COPY + HIDE as keys in a header row, and no ✕. audio just gained copy + ✕ (dotfiles =bd33440=).

Work: give bluetooth a copy key, give maint the overlay pair, and lift the four hand-rolled =_copy_output= implementations into one shared helper rather than a fifth copy. maint keeps HIDE alongside close, because its wall is a persistent session action log you collapse and keep, where net's, bt's, and audio's are per-run results you dismiss.

Copy text is per panel but one rule: it pastes as that panel's CLI prints, so the paste lines up with the terminal a user is already looking at. audio's =viewmodel.wall_copy_text()= is the worked example.

Consistent with the 2026-07-07 scope note on the sibling task above: net's compact glyph overlay is what standardizes, not maint's wide COPY-key header row.
** DONE [#C] Org-capture float popup grows too large :bug:hyprland:quick:solo:
CLOSED: [2026-07-14 Tue]
:PROPERTIES:
:LAST_REVIEWED: 2026-07-13
:END:
Craig answered the pre-flight (2026-07-14): cap at 120 wide, height proportional. Applied as 120 Emacs columns (11 px/col measured from the live daemon) = 1320 px wide, height 653 px from the old rule's aspect. Ratio's size rule shrank from the 1892x936 scratchpad match and both hosts gained a max_size growth cap (the field is max_size — bare "maxsize" is invalid and hyprctl reload won't say so; check hyprctl configerrors). Verified live: config clean, a probe window floats at exactly 1320x653. Dotfiles 9c4dc2f.
** DONE [#C] Panel smoke: faceplate state-word assertion fails on the live compositor :bug:dotfiles:test:
CLOSED: [2026-07-14 Tue]
Diagnosed and fixed within the session: not a race — test drift. Dotfiles b581d5d (2026-07-05) made the faceplate word the static subsystem identity (NETWORKING / BLUETOOTH / AUDIO) and updated the audio smoke, but the net and bt smokes kept asserting the retired live-state words and had failed on every run since. Both now assert the identity word like audio's (dotfiles 32cd99f); both smokes run RESULT: OK end to end, which also green-gates the doctor-streaming change.
** DONE [#C] Realtime lamp output for the net + bt doctors :feature:solo:
CLOSED: [2026-07-14 Tue]
:PROPERTIES:
:LAST_REVIEWED: 2026-07-09
:END:
Shipped in dotfiles 0318a91. Both doctors stream: diagnose() emits each step as it completes (bt streams the first diagnosis only — the fix loop's re-diagnoses would replay the chain), and a repair's row goes up amber at attempt start and settles green/red with narration + evidence at completion, so the lamp blinks for the repair's real duration. The 3.5-entry height cap turned out to already be in both wells (it landed with the doctor expansions), so only the streaming half needed building. 5 new tests across net + bt; both suites green; AT-SPI smokes at parity with HEAD (one pre-existing state-word failure, filed separately).
Retrofit the net doctor (=~/.dotfiles/net/src/net/doctor.py=) and bluetooth doctor (=~/.dotfiles/bluetooth/src/bt/doctor.py=) to stream results as a live output wall — one lamp per escalation step, amber while running, green on success, red on failure — instead of a final summary. Matches the maintenance-console doctor design (see [[file:docs/design/maintenance-console-design-ideas.org][maintenance-console-design-ideas.org]], "Doctor = live output wall"). Goal: every doctor in the system reads the same way. Both doctors already step through an escalation chain re-probing after each, so the steps are natural lamp boundaries.

Scope note (Craig, 2026-07-07): realtime lamp *behavior* only. The maintenance console's wider results-wall layout (date+time stamp column, COPY, persistent history) does NOT backport — the net/bt panels are ~400px wide and lack the horizontal real estate. Their existing output wells keep their compact layout; this task just makes them stream live.

Addendum (Craig, 2026-07-07): DO backport the 3.5-entry height convention — every panel's output well caps at 3.5 visible entries, the half-visible entry being the scroll cue, with the dark slate-on-black scrollbar. Layout stays compact per above; only the height cap + scroll affordance carries over.