Nomarchy/agent/JOURNAL.md

# Journal — append-only iteration log

One entry per iteration, **newest first**. This is the loop's short-term
memory: the next session reads the last 3–5 entries to orient. Keep an
entry under ~15 lines; durable lessons go to MEMORY.md, not here.

Template:

```markdown
## YYYY-MM-DD — <task title> (<commit sha, or "no commit">)
- **Task:** which BACKLOG item (or QA sweep / bootstrap / escalation)
- **Did:** what changed, in 2–4 lines
- **Verified:** tier reached (V0/V1/V2) + the actual commands/checks run
- **Pending:** V3 queued? follow-ups filed? anything [stuck]?
- **Next suggestion:** what the following iteration should probably take
```

---

## 2026-07-05 — launch-or-focus binds (iteration #39, item 17)
- **Task:** BACKLOG NEXT#17 (the UI review is Bernardo-gated; back to
  the queue head). Item deleted.
- **Did:** nomarchy.launchOrFocus (listOf submodule: key/mods/class/
  command/desc; opt-in, template example) → hyprland.nix generates
  binds through a shared nomarchy-focus-or-launch script (jq
  case-insensitive class match → focuswindow "(?i)" regex, else
  launch; uninstalled command → toast, so binds survive suite edits);
  rofi.nix renders the same entries into the SUPER+? cheatsheet via
  prettyKeys (single source). README row.
- **Verified:** V0; V1 — scratch downstream with 2 entries: both
  binds rendered (default command = lowercased class; custom
  mods/command/desc honored), tool bash -n + usage exit 64,
  cheatsheet rows present. Focus behaviour needs a session → V3
  queued.
- **Pending:** V3. Queue after this: #18 control center [big, needs
  the form-factor Decision], #19 Look&Feel menu (gated on ≥3
  entries), #20 [human], LATER tier.
- **Next suggestion:** a QA sweep (#18/#19/#20 are blocked-ish) — or
  item 19's precondition check (does Look & Feel earn 3 entries
  yet?).

## 2026-07-05 — placeholder dim + latte review + backdrop probe (iteration #38, item 28c)
- **Task:** item 28 slice (c) — the summer-rasi quickie + the light-
  theme capture.
- **Did:** both summer rasi: placeholder-color @bg0 → bg0 at 60%
  alpha (#…99), so prompt and placeholder no longer read as one
  phrase. Captured catppuccin-latte at 1080p: the generated menu is
  COHERENT on light themes (zebra rows on the 28b-retuned surface,
  hints dimmed) — closes the audit's "dark terminal on latte"
  concern as a stale preview. Backdrop suspicion probed with a
  dedicated VM (hyprctl getoption): misc:background_color IS set
  (0xFFEFF1F5) — awww's empty-state layer occludes it in the
  wallpaper-less VM; on hardware the fix covers the startup gap as
  intended.
- **Verified:** rasi edits grep-verified (rasi parses at rofi
  runtime; visual rides the next summer capture / V3); latte menu
  capture reviewed; probe output above.
- **Pending:** item 28 remaining: icon-weight `[human]`, hardware
  eyeballs (queued), identity-theme capture (optional), slice (d)
  needs nothing (parity verified).
- **Next suggestion:** item 28 is now Bernardo-gated (taste calls +
  V3); the loop should return to the general BACKLOG head (item 14
  watch / item 17 launch-or-focus).

## 2026-07-05 — rofi keybind hints (iteration #37, item 28c)
- **Task:** item 28 slice (c) — the hint column, the audit's biggest
  remaining UX item.
- **Did:** rofi.nix: menuHint builds a dimmed pango span
  (size small, alpha 55%) from keybinds.nix via the cheatsheet's
  prettyKeys — single source, hints can't drift from the live binds;
  applied to root (Apps/Theme/Power/Keybindings; Apps special-cased,
  it's rofi drun not a menu module), Tools (all eight leaves), System
  (Network/Bluetooth/DND); those three menus now pass -markup-rows.
  Contains-glob matching and the exact "↩ Back" match are unaffected
  (labels keep their prefix; Back stays markup-free).
- **Verified:** V0; V1 rendered menu bash -n + hint spans present +
  3× -markup-rows; VISUAL — tokyo-night capture: hints render dimmed
  and subordinate on normal AND selected (accent) rows; Tools/System
  rows correctly bare.
- **Pending:** slice-c remainder: summer-rasi placeholder dim,
  latte/identity captures, icon-weight `[human]`.
- **Next suggestion:** the placeholder-dim quickie + a latte capture
  in one tick; then item 28 pauses for Bernardo's icon-weight call
  and the hardware eyeballs.

## 2026-07-05 — bar group rhythm + @muted adoption (iteration #36, item 28c)
- **Task:** item 28 slice (c) — the two queued bar polish items.
- **Did:** waybar.nix generated CSS: margin-left 14px on the three
  right-cluster group heads (pulseaudio/powerprofile/battery — media
  · toggles · status; hidden heads degrade to uniform spacing);
  workspaces buttons, dnd bell, and muted volume moved from
  alpha(@text,.5) tints to the semantic @muted role — safe since 28b
  floor-gates muted/base ≥ 2.0 (the old comment warning against
  @muted documented the pre-28b world; rewritten to state the new
  invariant). Whole-swap bars untouched (authored identity).
- **Verified:** V0; V1 rendered CSS has 4 @muted uses + the margin
  rule; VISUAL — tokyo-night re-capture: themed backdrop confirmed
  (navy base, not black), group gap visible before the volume group,
  workspace pill legible. Cross-theme minima are guaranteed by the
  gated checker rather than per-theme eyeballs.
- **Pending:** slice-c remainder: rofi hint column, summer-rasi
  placeholder dim, icon-weight `[human]`, latte/identity captures.
- **Next suggestion:** rofi keybind-hint column (the biggest
  remaining UX win), or the two-line placeholder nit as a quickie.

## 2026-07-05 — summer-day capture review + themed backdrop (iteration #35, item 28c)
- **Task:** item 28 slice (c) — whole-swap capture review + the first
  cheap polish fix.
- **Did:** captured summer-day at 1080p (harness, VM): the authored
  slate menu + island-pill bar hold up with the 28b retunes —
  initially misread as polarity soup, but the rasi whole-swap
  deliberately mirrors the bar's slate strip: identity, keep (nit
  filed: undimmed entry placeholder). The surviving cross-theme fix:
  Hyprland `misc:background_color` = theme base (was compositor
  black — a glitch-looking flash behind missing/slow wallpaper,
  glaring on light themes). Punch list updated.
- **Verified:** V0; V1 — template-home rebuilt, rendered hyprland.conf
  carries `background_color=rgb(1a1b26)` (theme base). Backdrop
  visible confirmation rides the next capture run / V3.
- **Pending:** captures for latte + one identity theme; polish items
  queued: bar group margins, @muted adoption, rofi hint column, icon
  weight `[human]`.
- **Next suggestion:** bar polish slice (group margins + @muted) —
  code-level, verifiable by build + re-capture.

## 2026-07-05 — 1080p capture harness + first slice-c review (iteration #34, item 28c)
- **Task:** item 28 slice (c) prep — full-res captures for the
  per-surface polish pass.
- **Did:** tools/theme-shot.nix (maintainer tool, THEME env-driven):
  full themed desktop headless at 1920x1080 via the memory recipe +
  two fixes over the spike — home-manager.useGlobalPkgs (HM submodule
  otherwise misses the overlay → 'nomarchy-theme-sync missing') and a
  menu.png shot (hyprctl dispatch exec nomarchy-menu). Proven on
  tokyo-night; both shots read. Findings → item 28 punch list: rofi
  root width vs empty space (→ keybind-hint column idea), icon-weight
  mix (color Papirus vs mono bar) `[human]`, bar group margins
  confirmed, alpha-tint dim vs @muted role. Whole-swap module parity
  verified textually (no drift; cpu/memory omission is identity).
- **Verified:** harness runs end-to-end (exit 0, screenshots read);
  eval from tools/ with relative flake path OK. Wallpaper still
  doesn't paint (theme-sync needs ~/.nomarchy in the guest — fix
  noted in the header for the next run).
- **Pending:** capture summer-day/night (whole-swaps) + latte;
  then the polish slices; icon-weight call is Bernardo's.
- **Next suggestion:** run the retuned-theme captures + start the
  cheap polish items (group margins, @muted adoption).

## 2026-07-05 — P2 floors retuned + gated (iteration #33, item 28b done)
- **Task:** item 28 slice (b) remainder — the P2 contrast floors.
- **Did:** 18 hex retunes across 9 themes: muted raised on lumon/
  white/retro-82/everforest (OKLCH hue-preserving minimal solve to
  ≥2.05) + gruvbox→bg4 #7c6f64, nord→#616e88 (community standard),
  latte→overlay1 #8c8fa1; text-on-surface via surface nudges
  (ristretto darker, latte lighter); accentAlt darkened (latte,
  summer-day); warn darkened on the light themes (latte, summer-day,
  rose-pine-dawn) and flexoki-light's four statuses moved to the
  canonical 600 series. check-theme-contrast.py gates all four
  adopted floors (muted ≥2, text/surface ≥4.5, accentAlt ≥3, warn
  ≥2.5 — warn is the battery tint); good/bad stay audit-only
  (identity), so no exemption mechanism was needed. Slice (b) done.
- **Verified:** V0 + checks.theme-contrast GREEN (21×7); re-audit:
  contrast findings 20→1 (miasma bad = identity); flake check green.
  Rendering is V3 → queued (P2 eyeball entry).
- **Pending:** slices (c) per-surface polish (needs 1080p captures)
  and (d) whole-swap bar parity; ansi[] conventions decision still
  open (audit notes).
- **Next suggestion:** slice (c) prep — 1080p captures via the VM
  recipe, or hardware captures if Bernardo prefers eyeballing first.

## 2026-07-05 — P1 palette retunes + gated subtext floor (iteration #32, item 28b)
- **Task:** item 28 slice (b), P1s from the #31 audit.
- **Did:** summer-day: surface #5c6a72(=text!)→#efe7d0 raised cream,
  subtext #fdf6e3(=base!)→#6e828a slate; flexoki-light to canonical
  Flexoki tones: surface #E6E4D9(ui), overlay #B7B5AC, subtext
  #6F6E69(tx-2), muted #878580(base-500); miasma surface #000→#2e2e2e;
  kanagawa surface #090618→#2a2a37 (upstream sumiInk4 — old value was
  sumiInk0, darker than bg). check-theme-contrast.py gains the adopted
  `subtext/base ≥ 3` pairing (the item-25/27 trap, now impossible).
  Model note recorded: `overlay` is a mid-tone role, not a bg tier.
- **Verified:** V0 + checks.theme-contrast GREEN (21×3 pairings);
  re-audit: all four P1 findings gone (kanagawa now CVD-only findings);
  V1 — scratch downstream with summer-day ACTIVE builds, new surface
  hex present in rendered ghostty config. Looks are V3 → queued
  (four-theme eyeball).
- **Pending:** 28b remainder: P2 floors (muted ≥2, warn ≥2.5,
  text-on-surface ≥4.5, accentAlt ≥3) + exemption-list mechanism.
- **Next suggestion:** 28b P2s (7 muted retunes + 4 warn + 2
  text-on-surface + 2 accentAlt, then gate with exemptions).

## 2026-07-05 — theme design audit (iteration #31, item 28 slice a)
- **Task:** BACKLOG NEXT#28 slice (a) — the design-theory audit
  (human-requested; Bernardo asked to run it now).
- **Did:** tools/audit-theme-design.py (new, report-only): OKLCH
  lightness architecture, extended WCAG pairs, status-hue families,
  accent harmony, Machado protan/deutan simulation, ANSI slot
  semantics — over all 21 palettes; plus a visual pass over 4
  representative preview.png captures. Findings curated
  identity-aware into a P1–P3 punch list appended to item 28: 2
  invisible-subtext palettes (summer-day, flexoki-light), 2 inverted
  bg stacks (miasma, kanagawa), 7 muted-floor misses, warn-barely-
  reads ×4, a systemic "status is never color-only" rule (CVD
  collapse is inherent, glyph redundancy is the fix), explicit
  identity exemptions, light-theme ANSI convention decision, bar
  right-cluster grouping rhythm, latte dark-terminal verify.
- **Verified:** V0 (py_compile; report tool, not wired into checks).
  Analysis is palette-math + thumbnail previews; per-surface
  spacing/type audit needs 1080p captures (slice c prep).
- **Pending:** slice (b) — palette retunes + checker extension w/
  exemption list; needs Bernardo's eye on the P1 retunes (visual
  taste calls). Artwork needs none so far — no image-model ask yet.
- **Next suggestion:** slice (b) P1s (summer-day/flexoki subtext+
  surface, miasma/kanagawa stacks) — small hex edits, big payoff.

## 2026-07-05 — greeter theming from the JSON (iteration #30, item 16)
- **Task:** BACKLOG NEXT#16 — tuigreet themed from theme-state.json at
  system rebuild (the Plymouth model). Item deleted.
- **Did:** new modules/nixos/greeter.nix owns the greetd block (moved
  out of default.nix, one concern one file) + the theming: (1)
  console.colors = the theme's ansi[] (mkDefault; lands as
  vt.default_* kernel params — also themes raw ttys + LUKS prompt);
  (2) tuigreet --theme on NAMED slots (its parser is ratatui
  Color::from_str, verified in source; names hit the standard indexes
  the retinted VT palette now serves). Sparse state w/o ansi → skips
  cleanly.
- **Verified:** V0; V1 — template-system toplevel: baked greetd.toml
  carries the full --theme spec, kernel-params carry all three
  vt.default_{red,grn,blu}. A checks.greeter VM test was attempted and
  DROPPED: tuigreet dies under runNixOSTest even bare (harness limit,
  not our flag — bisected; nixpkgs tests use agreety) → MEMORY.md.
- **Pending:** V3 queued (rendering + theme-follows-rebuild).
- **Next suggestion:** #17 launch-or-focus, or the new #28 UI-review
  audit slice once Bernardo confirms its placement.

## 2026-07-05 — display profiles auto-switch (iteration #29, item 15b)
- **Task:** BACKLOG NEXT#15 slice b — opt-in hotplug auto-switch.
- **Did:** nomarchy-display-profile gains `match <outputs>` (exact set
  → else unambiguous largest subset → ties/none = exit 1);
  nomarchy-display-profile-watch polls `hyprctl monitors all` every 3s
  (exec-once, the keyboard-watcher pattern — session units race
  Hyprland IPC), reacts only to CHANGES, reads settings.displayProfileAuto
  LIVE each change (instant menu toggle, no rebuild) and applies via
  the tool, so auto-picks persist concrete profiles that rebuilds
  bake. Chose a separate boolean over an "auto" pseudo-profile for
  exactly that bake semantics. Menu: Auto-switch (on/off) row in
  Profiles. Template + README notes.
- **Verified:** V0; V1 — scratch 4-profile downstream builds; both
  scripts bash -n; match matrix exercised: exact ×3, unambiguous
  subset, tie → none, no-fit → none; watcher exec-once present; menu
  renders with the row. Real hotplug needs a session+dock → V3 (rides
  the slice-a queue entry).
- **Pending:** V3. Slice c (workspace binding) remains, tagged
  optional.
- **Next suggestion:** #16 greeter theming — item 15's remaining
  slice c is optional; a fresh item keeps variety.

## 2026-07-05 — display profiles slice a (iteration #28, item 15a)
- **Task:** BACKLOG NEXT#15 slice a, per the #27 design: option +
  overlay + applier + menu + template example.
- **Did:** nomarchy.displayProfiles (attrsOf (listOf monitorType));
  monitor logic extracted to pure modules/home/monitor-rules.nix
  (rule renderer + 3-layer resolve: profile replaces whole-by-name →
  resolution picks field-level with the disable-guard → wildcard),
  hyprland.nix consumes it and gains nomarchy-display-profile
  (list/active/apply/base; apply = baked hyprctl rules + state write
  --no-switch, base = clear + hyprctl reload) on PATH only when
  profiles exist; menu: Display gains a gated "Profiles ›" row + new
  display-profile mode (●/○ active mark, Base layout row); template
  example + README row. checks.display-profiles unit-tests resolve
  (incl. junk-state degradation) — the first HM-eval fixture attempt
  hit "path not valid" (eval store won't realise writeText/toFile
  under flake check), hence the pure-function extraction.
- **Verified:** V0 (check's asserts force at eval); V1 — scratch
  downstream through lib.mkFlake builds: tool bash -n + correct baked
  rules (disable, vrr, floats), hyprland.conf shows the exact overlay
  (wildcard + kept base + disabled panel + profile output), menu bash
  -n, list/usage smoke. Live switching needs a session+dock → V3.
- **Pending:** V3 queued. Slices b (hotplug auto-switch) and c
  (workspace binding) remain in the item.
- **Next suggestion:** #16 greeter theming, or 15b if momentum on
  displays is preferred.

## 2026-07-05 — display-profiles design pass (iteration #27, item 15, no code)
- **Task:** BACKLOG NEXT#15 — the entry itself demands "design first,
  write the plan into this entry before coding". Backlog-only iteration.
- **Did:** wrote the plan into item 15: `nomarchy.displayProfiles =
  attrsOf (listOf monitorType)` (reuse the existing submodule),
  `settings.displayProfile` state via the night-light pattern (instant
  hyprctl apply + bake on rebuild), whole-entry-by-name overlay with a
  disable-guard against stale resolution picks, self-gated menu row,
  three slices (option+overlay+menu / opt-in hotplug auto-switch /
  per-profile workspace binding), kanshi stays a non-goal.
- **Verified:** V0 (docs-only; flake untouched).
- **Pending:** nothing — 15a is now directly codeable.
- **Next suggestion:** 15a (option + overlay + menu picker), or #16
  greeter theming if a fresh session prefers a self-contained item.

## 2026-07-05 — scheduled lock-bump workflow (iteration #26, item 14 b+c)
- **Task:** BACKLOG NEXT#14 `[big]` — split: (a) checks-on-push shipped
  earlier; this ships (b) the weekly bump + (c) the fast-lane as
  workflow_dispatch on the same job.
- **Did:** .gitea/workflows/bump.yml — Mon 05:17 UTC cron + dispatch:
  nix flake update (pinned branches by construction) → gate `nix flake
  check --no-build` (py_compile/bash -n skipped: they don't read the
  lock) → commit --only flake.lock with the update's change list as
  body → push main (triggers check.yml as second net). Red gate/push
  race = failed run, nothing lands. Container recipe from check.yml.
- **Verified:** V0 (yq parse + bash -n each run block); logic simulated
  end-to-end in a scratch clone with a REAL nix flake update + bare
  remote: changed/unchanged branches, bystander file stays uncommitted,
  push lands; today's bumped lock evals green (Monday's baseline).
  Gitea 1.25.4 confirmed (schedule supported). The real cron pickup +
  token push are NOT confirmable from here → item 14 slimmed to
  "confirm first run".
- **Pending:** watch the first scheduled/dispatched bump run.
- **Next suggestion:** #16 greeter theming (next codeable item; #15
  needs its design pass written first).

## 2026-07-05 — color picker (iteration #25, item 13 final slice → item done)
- **Task:** BACKLOG NEXT#13 last slice — hyprpicker → clipboard from
  the menu + a direct bind. Item 13 complete; deleted from BACKLOG.
- **Did:** hyprpicker into systemPackages (next to grim/slurp); rofi.nix
  `colorpicker` mode (hyprpicker → wl-copy → hex-in-toast, Esc = silent
  cancel) + self-gated Tools › "Color picker" row + usage line;
  keybinds.nix SUPER+CTRL+P row (feeds bind + cheatsheet).
- **Verified:** V0; V1 — template-home AND template-system build;
  rendered nomarchy-menu bash -n OK with mode + gated row; cheatsheet
  txt has the SUPER+CTRL+P row; hyprland conf has the bind; system
  toplevel ships sw/bin/hyprpicker. Picking needs a compositor → V3.
- **Pending:** V3 queued (loupe → click → wl-paste round-trip).
- **Next suggestion:** #15 (display profiles) needs a design pass
  written into the entry first; #16 (greeter theming) is the next
  directly codeable item — or #14's bump workflow (eval-tier gate).

## 2026-07-05 — low-battery notifications (iteration #24, item 13 slice)
- **Task:** BACKLOG NEXT#13 second slice — the bar colors the battery
  at 25/10% but nothing *notified*.
- **Did:** pkgs/nomarchy-battery-notify (shellcheck-gated watcher:
  polls the same sysfs the bar reads — type=Battery, scope≠Device —
  one toast per downward crossing: normal@25, critical@10; re-armed by
  charging; self-gates on battery presence → desktop no-op) + overlay.
  HM unit in modules/home/battery-notify.nix behind
  nomarchy.batteryNotify.enable (default on), graphical-session-bound,
  libnotify on the unit PATH (notify-send resolved from PATH so the
  check can shim it). README row. Chose the tiny watcher over
  poweralertd: its UPower thresholds (20/5) wouldn't match the bar,
  and it toasts every plug/unplug.
- **Verified:** V0; V1 — template-home builds, unit present; V2 GREEN —
  new checks.battery-notify (test_power fake battery): self-gate exits
  0 without a battery, silence at 80%, exactly one low toast crossing
  25, one critical-urgency toast crossing 10, charging re-arms → a
  second drain toasts again.
- **Pending:** V3 queued (real toast through swaync while draining).
- **Next suggestion:** item 13 final slice — hyprpicker color picker
  (Tools › entry + SUPER+CTRL bind).

## 2026-07-04 — caffeine/idle-inhibit toggle (iteration #23, item 13 slice)
- **Task:** BACKLOG NEXT#13 first slice — idle_inhibitor in the
  generated bar (summer-night had it; reverse parity gap).
- **Did:** waybar.nix: idle_inhibitor module (󰾪/󰅶, tooltips, placed
  after the ⏺), CSS membership + `.activated { color: @warn; }`.
  summer-day: module + CSS mirrored from summer-night (same glyphs,
  no tint — glyph change is the state signal there). keybinds.nix
  extra row → cheatsheet mentions the bar click.
- **Verified:** V0; V1 — template-home built; rendered config has the
  module second in modules-right with correct glyphs; CSS rules
  present; cheatsheet row renders. Inhibit behaviour needs a session
  → V3 queued (hold past the lock timeout).
- **Pending:** V3. Item 13 remaining slices: low-battery
  notifications, color picker.
- **Next suggestion:** item 13 next slice (low-battery notifications,
  gate on power.laptop) — or #14 lock-bump CI if variety is better.

## 2026-07-04 — screen recording in Capture (iteration #22, item 12)
- **Task:** BACKLOG NEXT#12 — record region/screen from the menu, stop
  from the bar.
- **Did:** modules/home/recording.nix: `nomarchy-record`
  start/stop/active/status — wl-screenrec first, wf-recorder fallback
  if it dies at startup (no-VAAPI case), SIGINT-finalized stop,
  runtime pidfile, signal-8 bar pokes. Capture menu: 4 record rows that
  self-swap to "■ Stop recording" while one runs. Waybar: self-gating
  custom/recording (⏺ REC, @bad red), click = stop — the one stop
  surface. Parity: both summer whole-swap bars got the module + CSS
  (@red). DEVIATIONS: saves to ~/Videos/Recordings (an .mp4 in
  Pictures/Screenshots felt wrong — same timestamp naming though);
  audio = two extra explicit rows, not a toggle state.
- **Verified:** V0; V1 — template-home builds; rendered bar config has
  the module (first in modules-right) + CSS rule; summer jsonc still
  parse; nomarchy-record dry-run: status empty+exit 0 idle, active
  exit 1, usage exit 2; both recorders in home-path. Real capture
  needs a session → V3 (headless GL recipe can't confirm an encoder).
- **Pending:** V3 recording exercise (incl. VAAPI-vs-fallback check).
- **Next suggestion:** NEXT#13 niceties, one slice: idle-inhibit
  (caffeine) toggle — reverse-parity gap already noted in the item.

## 2026-07-04 — state-file validation & friendly errors (iteration #21, item 11)
- **Task:** BACKLOG NEXT#11 — hand-edited state must fail with field +
  problem + fix, never a Nix stack.
- **Did:** (a) tool: validate_state() (appearance schema = hard errors;
  settings.*/unknown keys = warnings), `validate` subcommand,
  validate-BEFORE-write in write_state (invalid set/apply never touches
  disk), JSON syntax errors now say line/column/fix. (b) theme.nix:
  same schema at eval time on `parsed` (post-defaults, so sparse files
  stay fine); throws a multi-line message naming each field, the got-
  value, and the three fix paths. Doctor's JSON fix-text now points at
  `validate`.
- **Verified:** V0; V1 — new checks.theme-sync-validate GREEN: good
  template passes, trailing comma → line/col message, bad hex/type →
  field named, unknown key → warning only, invalid `set` refused with
  the file byte-identical (cmp). Negative eval of a broken scratch
  downstream through mkFlake shows the friendly throw (both fields);
  good-path template-home still builds.
- **Pending:** nothing — no hardware dependency.
- **Next suggestion:** NEXT#12 (screen recording) or #13 niceties
  slice; #12 first (it reshapes the Capture submenu the OCR LATER item
  waits on).

## 2026-07-04 — nomarchy-doctor (iteration #20, item 10)
- **Task:** BACKLOG NEXT#10 — one-shot read-only health check.
- **Did:** pkgs/nomarchy-doctor (writeShellApplication → shellcheck-
  gated; a package so the VM check runs on a minimal node): failed
  system+user units (user bus self-skips), disk space on real
  filesystems only (fstype allowlist skips tmpfs/9p → VM-safe; device
  dedupe), state-file parses + git-tracked, flake dirty/behind
  (warn-level), generation age via the profile SYMLINK mtime (store
  paths are epoch-1), snapper timer when enabled. Every ✖ prints its
  fix; exit 1 on any ✖. Wired: overlay, systemPackages, System ›
  Doctor menu row (terminal), README §5.
- **Verified:** V0; V1 package build (shellcheck) + real-hardware
  smoke: correctly flagged a genuinely failed user unit on the dev box;
  V2 GREEN — VM test: induced failed unit → exit 1 + names it + prints
  fix; reset-failed → healthy exit 0. Menu rebuild bash -n ok.
- **Pending:** V3 menu-row check queued. Waybar-warning follow-up →
  LATER.
- **Next suggestion:** NEXT#11 (state-file validation) — doctor's
  "parses" check is its little sibling; 11 adds validate-before-write
  + eval-time schema messages.

## 2026-07-04 — System › Rollback menu (iteration #19, item 9b → item 9 done)
- **Task:** BACKLOG NEXT#9 half (b) — undo one menu away.
- **Did:** new `rollback` case in nomarchy-menu (System submenu row):
  lists the last 10 `home-manager generations` (newest marked current),
  picking one activates it in a terminal — HM's supported rollback,
  reversible, so no typed-yes. System-level undo LINKS OUT by design:
  Snapshots row → existing flow; "boot an older generation (how)" →
  instruction notification pointing at docs/RECOVERY.md §3 (no
  destructive reimplementation in rofi). RECOVERY §1 cross-references
  the menu path. Item 9 complete → deleted.
- **Verified:** V0; V1 — template-home built, generated nomarchy-menu
  bash -n clean, row-render + pick-parse logic run against a fixture
  `generations` output (id/path extraction correct). V2 skipped
  deliberately: the only novel logic is the parsing (fixture-covered);
  the activate step is HM's own mechanism. V3 queued.
- **Pending:** V3 rollback-menu exercise on hardware.
- **Next suggestion:** NEXT#10 `nomarchy-doctor` (V2-able in a VM).

## 2026-07-04 — sys-update/-rebuild what-changed diff (iteration #18, item 9a)
- **Task:** BACKLOG NEXT#9 half (a) — a human diff after system rebuilds.
- **Did:** both sys-update and sys-rebuild (twins stay twins) capture
  /run/current-system before, and after the switch print
  `nvd diff before after` (store-path-pinned ${"$"}{pkgs.nvd}, `|| true`
  so it can't fail a succeeded rebuild; identical-path fast-path says
  "no changes"). README §3 shortcut block notes it.
- **Verified:** V0; V1 — template-system toplevel rebuilt, generated
  scripts extracted + bash -n, nvd smoke-run against the real closure
  (correct output format). Runtime behaviour is a trivial shell path —
  no VM needed.
- **Pending:** half (b) — the System → Rollback menu flow (desktop
  generations picker; system → boot menu + snapshots pointer).
- **Next suggestion:** item 9b — it completes the item; V2 the
  generation-activate path.

## 2026-07-04 — viewers + mime defaults (iteration #17, item 8)
- **Task:** BACKLOG NEXT#8 — PDF/image viewers + xdg-mime defaults.
- **Did:** modules/home/viewers.nix (`nomarchy.viewers.enable`, default
  on): programs.zathura (Stylix themes it — target added to stylix.nix's
  explicit list, autoEnable=false bit us first try) + imv.
  modules/home/mime.nix (`nomarchy.mime.enable`): mkDefault
  xdg.mimeApps for pdf→zathura, image/*→imv, av→mpv, text→code,
  dir→thunar, html/http(s)→firefox as INERT entries (browser Decision
  untouched — absent .desktop entries are skipped by GIO). README rows
  added (option-docs enforces them now). DEVIATION from the item text:
  viewers are a module toggle, not template packages — zathura's
  theming needs its HM module, which is exactly the "real config"
  boundary of the toggle discipline.
- **Verified:** V0; V1 — template-home generation builds; asserted
  rendered mimeapps.list (all associations), themed zathurarc
  (Tokyo Night rgb), imv/zathura in home-path. V3 smoke queued.
- **Pending:** V3 open-a-file smoke; browser Decision now smaller
  (inert entries ship; only "active in suite?" remains).
- **Next suggestion:** NEXT#9 (update & rollback UX) — or #10 doctor.

## 2026-07-04 — docs slice (c): install story — item 6 COMPLETE (iteration #16)
- **Task:** BACKLOG NOW#6 slice (c), the last one — install/first-run
  narrative end to end.
- **Did:** read README §2 (ISO/installer) → §3 (downstream) → template
  README as one story; checked the facts en route: mkFlake's arg
  surface == the §3 table (src/username/hardwareProfile/system, lib.nix),
  disko really makes @snapshots + the hibernation swapfile, .#default /
  .#me names consistent. One seam fixed: §3 showed `nix flake init -t`
  but never handed off to the template README's six first-run steps —
  connective paragraph added (+ "the installer does all of this").
  Item 6 done → deleted from BACKLOG (NOW now empty); ROADMAP § Full
  docs review flipped to ✓ with the slice summary.
- **Verified:** V0 (docs-only) — flake check clean pre-commit.
- **Pending:** docs-site-vs-Markdown stays a Decision.
- **Next suggestion:** NEXT#8 (viewers + mime defaults) is the new queue
  head; note its "Default browser" Decision dependency — the mime module
  half can proceed with mkDefault degradation regardless.

## 2026-07-04 — docs slice (d): recovery runbook (iteration #15, item 6d)
- **Task:** BACKLOG NOW#6 slice (d) — write the undo story.
- **Did:** docs/RECOVERY.md, symptom-ordered: (1) bad theme/HM change →
  generations / apply-known-good / git-revert ~/.nomarchy; (2) desktop
  won't start → Ctrl+Alt+F2, greetd + user journals, VM-GL and
  preactivate-log pointers; (3) system change → systemd-boot older
  generation (kept: 10) + make-it-stick via sys-rebuild; (4) files →
  btrfs-assistant / sudo nomarchy-snapshots (undochange, root rollback,
  typed-yes); (5) ISO + nixos-enter. Every command checked against the
  tree (greetd cfg, configurationLimit, snapshots script actions,
  helper names). Linked from README tree + §5.
- **Verified:** V0 (docs-only) — flake check clean pre-commit.
- **Pending:** slice (c) install-story read-through — the last of item 6.
- **Next suggestion:** finish item 6 with slice (c).

## 2026-07-04 — docs slice (b): drift pass (iteration #14, item 6b)
- **Task:** BACKLOG NOW#6 slice (b) — OVERRIDES.md / TESTING.md /
  templates/downstream/README.md vs the live tree.
- **Did:** spot-checked every factual claim: mkDefault sites (hyprland
  behaviour knobs, ghostty paddings, waybar mainBar/style), bind-list
  normal priority, `set ui.gapsOut` CLI key, awww(-daemon) naming,
  21 presets, tokyo-night's 4 wallpapers, summer-night's light @fg bar,
  live `,highres` mkForce, .gitea/workflows/check.yml, QEMU gl flags,
  template `.#me` + ~/.nomarchy/NOMARCHY_PATH. All accurate. Fixed the
  two real gaps: OVERRIDES §2 predated `nomarchy.monitors` (now routes
  there + quick-ref row); TESTING §1 lacked the shell-syntax check §1b
  claims CI runs (added `bash -n`).
- **Verified:** V0 (docs-only; flake untouched — check run pre-commit).
- **Pending:** slices (c) install story, (d) recovery runbook.
- **Next suggestion:** item 6 slice (c) or (d) — (d) is self-contained
  and ships user value (the runbook).

## 2026-07-04 — docs slice (a): option-surface reconcile (iteration #13, item 6a)
- **Task:** BACKLOG NOW#6, first slice — README tables vs live options.
- **Did:** split item 6 into slices (a–d) in BACKLOG. Reconciled: 9
  undocumented options (autoTimezone.enable!, camera.hideIrSensor +
  irMatch!, intel.guc, amd.pstate/vaapi, package, system.stateFile,
  restic.paths) — rows/inline mentions added, no stale rows found.
  Made it permanent: `checks.option-docs` — flake.nix eval-walks option
  names from the 4 option files (config halves stay lazy; dummy args)
  → tools/check-option-docs.py enforces row-or-`.leaf`-mention + flags
  stale rows.
- **Verified:** V0; V1 — check builds green (78 options); negative test:
  same script vs the pre-fix README reports exactly the 9 gaps.
- **Pending:** slices (b) doc drift pass, (c) install story, (d)
  recovery runbook.
- **Next suggestion:** item 6 slice (b) — or NEXT#8 (viewers+mime) for
  variety; both fine.

## 2026-07-04 — Back audit (iteration #12, item 24)
- **Task:** BACKLOG NOW#24 — every list menu ends in ↩ Back.
- **Did:** audited every `rofi -dmenu` call site in the generated
  nomarchy-menu + nomarchy-vpn. Already compliant: power, power-profile,
  theme, clipboard, files, audio, display (both stages), capture,
  tools, system, VPN main/import/exit-node/tailscale. Fixed: keybinds
  cheatsheet (Enter was a silent exit) now ends in ↩ Back → root
  picker. Documented the exceptions in rofi.nix: external modi
  (calc/emoji/networkmanager_dmenu/rofi-pulse-select) can't take an
  injected row (Esc = back); free-text prompts (web/ask) have no list.
- **Verified:** V0; V1 — built nomarchy-menu, bash -n OK; awk audit
  over the generated script: only rows lacking back are the root
  picker (by design) and the free-text prompt. V3 spot-check queued.
- **Pending:** V3 spot-check. NOW now holds only item 6 (docs, [big]).
- **Next suggestion:** item 6 first slice, or NEXT#8 (viewers+mime).

## 2026-07-04 — network menu blank rows (iteration #11, item 22)
- **Task:** BACKLOG NOW#22 — nameless entries in the network picker.
- **Did:** read the pinned networkmanager_dmenu 2.6.2 source: with
  `compact = False` (our config!) it inserts literal empty rows as
  section separators (eth/wifi/vpn) — the systematic blank entries.
  Flipped config.ini to `compact = True` (also unpadded row format,
  better in rofi's proportional font). Hidden-SSID APs additionally
  render ONE nameless sec+bars row (dedup keeps one; not config-
  filterable) — documented, V3-gated, source-patch follow-up if seen.
- **Verified:** V0; V1 — rendered config.ini shows compact = True.
  Behavior needs Wi-Fi hardware → V3 queued.
- **Pending:** V3 on the Latitude (no separators; hidden-AP residual?).
- **Next suggestion:** item 24 (Back audit) — grep-the-generated-script.

## 2026-07-04 — waybar/rofi palette-safe pairings (iteration #10, item 27)
- **Task:** BACKLOG NOW#27 — generated waybar/rofi CSS invisible on
  on-surface palettes (flexoki-light runs it; summers whole-swap).
- **Did:** waybar.nix generated CSS: module row + #window →
  alpha(@text, 0.85); inactive workspaces / dnd / pulseaudio.muted →
  alpha(@text, 0.5) (gruvbox's muted/base was 1.27). rofi.nix: @dim →
  text+80, @surface → text+1a (#RRGGBBAA fg tints). Contrast check
  docstring/pairings now cover all three generated surfaces. Status
  glyphs (good/warn/bad on base, worst 2.05) left as-is — palette
  design, human call. Generated bar only; whole-swaps untouched.
- **Verified:** V0; V1 — checks.theme-contrast green (21 themes),
  downstream-template-home builds, rendered waybar style has zero
  @subtext/@muted/@surface, rendered custom.rasi shows fg-derived
  tints. NOT visually verified — V3 queued (flexoki-light eyeball).
- **Pending:** V3 flexoki-light + dark-theme regression spot-check.
- **Next suggestion:** item 22 (network menu rows) — V1-actionable.

## 2026-07-04 — swaync contrast fix + theme-contrast check (iteration #9, item 25)
- **Task:** BACKLOG NOW#25 — swaync text invisible on summer-day.
- **Did:** swaync.nix now uses only palette-safe pairings: body/summary
  = @text on @base (worst ratio 5.18 across all 21 themes; subtext/base
  was 1.00 on summer-day + flexoki-light), hover rows + widget buttons
  = alpha(@text, 0.1) tints (contrast by construction); dropped the
  subtext/surface defines. NEW permanent check `checks.theme-contrast`
  (tools/check-theme-contrast.py) asserts the hex-on-hex text pairings
  across every theme JSON. Audit found the same mispairing in generated
  waybar/rofi CSS, live on flexoki-light (no whole-swap) → NOW item 27.
- **Verified:** V0; V1 — checks.theme-contrast builds green (21 themes),
  rendered services.swaync.style contains zero @subtext/@surface.
  NOT V2/V3-verified visually — queued (Bernardo is at the Latitude).
- **Pending:** V3 notify-send readability check on summer-day; item 27.
- **Next suggestion:** item 27 (same bug class, waybar/rofi surfaces).

## 2026-07-04 — SUPER+? bind fix (iteration #8, item 26)
- **Task:** BACKLOG NOW#26 — cheatsheet bind never fires on hardware.
- **Did:** keybinds.nix: `$mod, question` → `$mod SHIFT, question`
  (? is Shift+/, Hyprland matches the exact modmask); rofi.nix
  prettyKeys collapses "SHIFT + ?" → "?" so the row stays SUPER + ?.
  Gotcha added to MEMORY.md (shifted keysyms need SHIFT in mods).
- **Verified:** V0 (flake check --no-build); V1 both surfaces — eval'd
  hyprland settings.bind renders `$mod SHIFT, question, exec,
  nomarchy-menu keybinds`, and built nomarchy-menu's cheatsheet file
  renders `SUPER + ?`. No other shifted keysyms in the bind list.
- **Pending:** V3 queued (SUPER+? on the Latitude after home-update +
  relogin).
- **Next suggestion:** item 25 (swaync contrast on summer-day).

## 2026-07-04 — Latitude hardware sweep #2 (Bernardo; results booked, no code)
- **Task:** V3 sweep on the Latitude after pulling ≥18b8545 + reboot.
- **PASS:** waybar supervisor (theme switches clean; `pkill -x waybar`
  respawns instantly); snapshots — themed polkit prompt → GUI as root,
  fzf fallback works (restore/rollback gate not exercised → residual
  queued); `sys-rebuild`; night-light enable/disable; auto-commit
  (theme commits + the toggle self-commit); hibernate single-unlock +
  suspend-still-locks (1b0eeea CONFIRMED).
- **PASS (late confirm):** auto-timezone — `timedatectl` shows
  Europe/London.
- **Bug:** SUPER+? never fires — `$mod, question` misses because a real
  ? carries SHIFT in the modmask → BACKLOG 26 (bind-only: the menu
  itself opens fine from a terminal). SUPER+SHIFT+K no-op was the
  comma-layout gate working (none configured) — retest queued.
- **Next suggestion:** item 25 (swaync contrast) or 26 (cheatsheet bind).

## 2026-07-04 — Waybar resilience + sys-rebuild (iteration #7, items 21+23)
- **Task:** the two top Latitude findings.
- **Did:** (21) `nomarchy-waybar` supervisor (waybar.nix, exec-once'd
  from hyprland.nix): ANY waybar exit respawns, crash-loop guard (5
  fast exits → critical notify + stop), TERM trap for a real stop.
  theme-sync now prefers a clean `pkill -x waybar` when the supervisor
  is running (restart with fresh config+style — avoids waybar's
  in-place SIGUSR2 reload entirely, the suspected double-reload race
  with reload_style_on_change during symlink flips); SIGUSR2 stays as
  the unsupervised fallback; reload_style_on_change kept for manual
  home-update restyles. (23) `sys-rebuild` — snapshot-first rebuild
  against the current lock, no `nix flake update`; README §3/§5 + motd.
- **Verified:** V0; V1 (HM renders `exec-once=nomarchy-waybar`,
  supervisor bash -n; `sys-rebuild` in systemPackages). V2: headless
  software-GL desktop VM (recipe from memory — gotchas hit: node needs
  the overlay, users.users.<u>, useGlobalPkgs) — SIGKILL→new pid, clean
  kill→respawn, SIGUSR2→alive. Result recorded below once green.
- **Pending:** V3 on the Latitude: theme switch survives / bar returns.
- **Next suggestion:** item 22 (network menu rows) or 24 (Back audit).

## 2026-07-04 — Keyboard cycle bind + parity (iteration #6) + Latitude triage
- **Task:** BACKLOG NOW#5; mid-iteration Bernardo delivered Latitude
  5410 hardware-QA findings (the machine got a real session!).
- **Did:** `multiLayoutBinds` in keybinds.nix (SUPER+SHIFT+K →
  `switchxkblayout current next`), gated in hyprland.nix AND the rofi
  cheatsheet on the same comma-in-layout condition; `hyprland/language`
  + `#language` CSS added to both summer whole-swaps (shows even
  single-layout there — static JSON can't gate; deliberate). Reverse
  parity gap found (summer-night has idle_inhibitor, generated bar
  doesn't) → folded into item 13. **Triage:** four new NOW items from
  hardware: #21 Waybar crash on theme switch + no respawn (top),
  #22 nameless network-menu entries, #23 `sys-rebuild` (no-update
  rebuild), #24 Back-everywhere audit.
- **Verified:** V0 (JSON + flake check) + V1 both directions
  (single-layout build: no bind; `us,de` via extendModules: bind + the
  cheatsheet row render; summer JSONs parse).
- **Pending:** V3 cycle-bind check queued; items 21/22/24 carry V3
  re-checks on the Latitude.
- **Next suggestion:** #21 (waybar crash — stability pillar, user-hit).

## 2026-07-04 — Webcam follow-ups (loop iteration #5, quick item)
- **Task:** BACKLOG NOW#7 (Bernardo asked for "a quick item").
- **Did:** template `home.packages` gains a commented "Webcam tuning"
  pair (cameractrls GUI, v4l-utils CLI — bare packages, so template not
  toggle, per the option-surface rule); the portal/Flatpak libcamera IR
  gap is written up as a PROPOSED item with three options (recommend:
  document-only now, libcamera-rule investigation when the T14s is
  available). CI runs #60/#61 confirmed green (earlyoom + snapshot fix).
- **Verified:** V0 (flake check — template evals through mkFlake;
  changes are comments + agent docs).
- **Next suggestion:** NOW#5 (keyboard cycle bind + summer parity) —
  the last code item in NOW besides the docs pass.

## 2026-07-04 — Snapshot GUI un-broken + polkit agent (loop iteration #4)
- **Task:** BACKLOG NOW#4 — btrfs-assistant segfault.
- **Did:** re-diagnosed instead of overriding. gdb: crash is in
  libbtrfsutil's *unprivileged* subvolume iteration (btrfs-progs 6.17.1;
  upstream fix `886571653` post-6.17.1; symbol versions verified — not
  ABI drift). VM A/B: root works (exit 0), user crashes (139). The
  launcher runs it as root → GUI fine *if* pkexec can prompt — and the
  distro shipped **no polkit agent** (all pkexec silently failed; the
  actual root cause of "GUI is dead"). Shipped: hyprpolkitagent
  (exec-once, Stylix-themed Qt), menu re-pointed to the launcher with
  the fzf flow as fallback, `checks.snapshot-gui` canary (root path +
  offscreen event loop on real btrfs). No btrfs-progs patch — root-side
  flows don't hit the bug; the fix arrives with a lock bump.
- **Verified:** V0 + V2 (snapshot-gui check, run for real) + V1 (HM
  generation renders the agent exec-once; menu bash -n).
- **Pending:** V3 queued — first on-hardware themed polkit prompt.
- **Next suggestion:** NOW#5 (keyboard cycle bind + summer parity).

## 2026-07-04 — CI is green (run #58) — item 20 main half closed
- **Task:** monitor follow-up, no code. Run #58 (the .gitea path fix)
  **succeeded** — first green CI: the container recipe works on the
  live runner, eval tier confirmed end to end.
- **Did:** BACKLOG item 20 reduced to its stretch (KVM runner → enable
  vm-checks), moved to NEXT. Run #59 (earlyoom push) monitored.
- **Next suggestion:** NOW#4 (btrfs-assistant override attempt).

## 2026-07-04 — Memory-pressure protection shipped (loop iteration #3)
- **Task:** BACKLOG NOW#3 — default-on userspace OOM killer.
- **Did:** `modules/nixos/oom.nix`: earlyoom (mkDefault on, desktop
  notifications, session-plumbing `--avoid` list, no `--prefer`) +
  `systemd.oomd.enable = false`. Choice rationale: Hyprland session is
  one cgroup scope → oomd's cgroup-level kill would nuke the desktop;
  earlyoom is process-level. Found en route: nixpkgs default-enables
  oomd *inert* (enable=true, all slice toggles false). README "beyond
  the surface" note added.
- **Verified:** V0 (flake check) + **V2** — new `checks.oom-protection`
  runNixOSTest, run for real: hog (chunked allocator, 686 MB peak in a
  1 GB VM) SIGTERM'd by earlyoom in 0.1 s, bystander unit survived,
  oomd asserted inactive. 25 s test, cheap enough for the suite.
- **Pending:** nothing hardware-specific (behavior fully VM-provable).
- **Next suggestion:** NOW#4 (btrfs-assistant override attempt). Also:
  CI run #58 (the .gitea fix) was *running* at commit time — check its
  outcome next tick; this push triggers #59.

## 2026-07-04 — CI fix: wrong workflow dir (correction to iteration #2)
- **Task:** Bernardo reported no run despite an active runner + shared
  his compose file: the server is **Gitea** (gitea/act_runner), not
  Forgejo — I'd shipped the workflow to `.forgejo/workflows/`, which
  Gitea never reads. The legacy repo's `.gitea/workflows/` path was the
  clue I under-weighted.
- **Did:** `git mv` → `.gitea/workflows/check.yml`; corrected the
  Forgejo references in the workflow header, TESTING.md §1b, BACKLOG
  item 20 (now: watch the first run; runner was never the problem);
  MEMORY.md gains the Gitea-not-Forgejo line.
- **Verified:** the push of this commit is itself the test — the run
  must appear; monitoring.
- **Pending:** first green run (item 20); runner-container memory
  headroom is the watch-out.

## 2026-07-04 — CI checks-on-push shipped (loop iteration #2)
- **Task:** BACKLOG NOW#2 — Forgejo Actions workflow.
- **Did:** `.forgejo/workflows/check.yml` (push to main/v1 + dispatch):
  flake check --no-build, theme-sync py_compile, bash -n over tracked
  .sh. Scoped to the **eval tier** deliberately — API probing +
  archaeology of the legacy `.gitea/workflows/check.yml` (57 runs on
  this instance) showed the runner is an act_runner docker container
  (no KVM/systemd), so the VM suite can't run there; a commented
  `vm-checks` job documents the KVM-runner upgrade path. Legacy's
  hard-won container gotchas (nixbld setup, sandbox=false for Stylix
  IFD, Nix 2.31.5 pin) carried over verbatim. TESTING.md §1b added.
- **Verified:** V0 locally (same commands the workflow runs, minus the
  container Nix install) + yq YAML parse. The real green run needs the
  runner to be alive — **not confirmable from here** (runners API is
  401 unauthenticated); watching the run after push.
- **Pending:** new `[human]` item 20 — confirm/re-register the runner;
  stretch: a KVM runner unlocks the vm-checks job.
- **Next suggestion:** NOW#3 (memory-pressure protection).

## 2026-07-04 — Opt-in auto-commit shipped (loop iteration #1)
- **Task:** BACKLOG NOW#1 — auto-commit of state mutations (in-flake
  state Phase 4). First autonomous /loop iteration.
- **Did:** `settings.autoCommit` live-read by nomarchy-theme-sync;
  pathspec-limited commit of theme-state.json on apply/set (fires if flag
  on before OR after the write → the off-toggle commits too); identity
  fallback; same-value no-op; `bg` excluded. Menu: System › Auto-commit
  (instant, self-gated on `.git`). Rider: `get` prints booleans as JSON
  `true`/`false`, fixing the stuck "Auto timezone (on/off)" menu label.
  BACKLOG numbering switched to stable IDs (no renumbering).
- **Verified:** V0 (py_compile, flake check) + V1 (HM generation builds;
  generated menu `bash -n`) + a 7-assertion sandbox git-repo round-trip
  (enable/set/same-value/disable/post-disable/apply/identity-fallback).
- **Pending:** V3 queued in HARDWARE-QUEUE (on-machine toggle + apply).
- **Next suggestion:** NOW#2 (CI checks-on-push) — but its runner half
  may need Bernardo to register a Forgejo runner; deliver the workflow
  regardless.

## 2026-07-04 — Backlog revision pass (human-requested)
- **Task:** Bernardo asked for a philosophy-aligned revision of BACKLOG.md
  ("I'm sure I'm forgetting important items").
- **Did:** verified guessed gaps against the tree first (GC/optimise/boot
  limit + all 21 previews already ship — dropped those ideas). Added the
  confirmed gaps: CI-on-push promoted to NOW#2 (agents push to main now),
  memory-pressure protection NOW#3 (nothing mitigates OOM; the release
  bump died to one), viewers+mime defaults #8, update/rollback UX #9,
  nomarchy-doctor #10, state-file validation #11, screen recording #12,
  niceties batch #13 (idle-inhibit, low-battery notify, hyprpicker), OCR
  in LATER; recovery runbook folded into the docs item; zram + default
  browser added to Decisions.
- **Verified:** V0 (docs-only).
- **Pending:** Bernardo to sanity-check the new NOW ordering, esp. CI at
  #2 and the two new Decisions.
- **Next suggestion:** unchanged — NOW#1 (opt-in auto-commit).

## 2026-07-04 — Bootstrap the loop infrastructure (this commit)
- **Task:** create `agent/` (human-requested): LOOP/GOALS/BACKLOG/
  CONVENTIONS/MEMORY/HARDWARE-QUEUE/JOURNAL, root CLAUDE.md, README +
  ROADMAP pointers.
- **Did:** reworked the forward half of docs/ROADMAP.md into the tiered
  BACKLOG (5 NOW / 6 NEXT / LATER / Decisions); collected every pending
  on-hardware check into HARDWARE-QUEUE.md; seeded MEMORY.md from the
  ROADMAP's decision records.
- **Verified:** V0 (docs-only change; `nix flake check --no-build` run to
  confirm the tree still evaluates).
- **Pending:** Bernardo to sanity-read GOALS.md pillars + BACKLOG tiering
  — reorder freely, the order *is* the instruction.
- **Next suggestion:** BACKLOG NOW#1 (opt-in auto-commit — Phase 4 of
  in-flake state, already the agreed next step).