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:

## 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

---

2026-07-04 — 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., 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).