Docs 02/05/07/13 and capability F12 (PROVEN-LIVE); the audit's C and D sections; R-1020 (the
duration reads 0) and R-1021 (the hub build builds a stale checkout) filed; STATUS one screen.
Register 133 -> 140 (this session: opened 2, closed 1; another session added 6). Both plugs ON.
Report: REPORT-immich-repair-2026-10-14.md.
Claude-Session: https://claude.ai/code/session_012qRErfCoiTkvDK9N5XHbzb
Two test drives enrolled through the dashboard; no test media in any backup tier (A6, three proofs); Tester 1 follows a
credential-free local mirror of the drill catalog (R-992 worked around a third way). The live-fix task may start.
The operator's half of the repair after an unclean boot (controller 0.311.0): counts only, never a
file name; warning mails, info is recorded. Registered in allowedEventTypes and operatorOnlyEvents in
the same commit, plus the per-app cooldown list; tests pin the pair, a household that enabled the type
gets nothing, and the counts line (red-proofed). Decision 236 (operator, 2026-10-14): the box asks
Immich through Immich's own job queue. R-1007 updated; the scripts' run on 9202 added as evidence.
Claude-Session: https://claude.ai/code/session_012qRErfCoiTkvDK9N5XHbzb
Read-back: live sizing-data.json?v=4 byte-equal to the commit; the presets keep their verdicts,
Immich alone on 8 GB still tight; only the set without Immich changed. Spike B2 on 9202:
POST /api/assets/jobs regenerate-thumbnail rebuilt exactly the 10 broken assets (1.55 s, 5.3 s CPU),
0 of the other 20 photos' 40 files rewritten (control: all rewrote 40 of 40); an empty file maps to
its asset by the first 36 characters of its name. Immich removed through the product, no leftovers.
Register 134 -> 133 (opened 0, closed 1). STATUS D9/D10 ruled.
Claude-Session: https://claude.ai/code/session_012qRErfCoiTkvDK9N5XHbzb
demo-hp 12:06:52 UTC, 1.53 GB written: at boot the dead task was judged dead, the partial and its .tmp
removed by the root helper, the next backup finished (5.65 GB), host root back below its pre-cut use.
Orphan thin device 4 repaired first (operator B2 - A); R-1012 P2 -> P3, detection stays open.
Delivered by signed jobs to Tester 1 and the N100 (probe 71/71, real sudo checked). Controller 0.310.0
everywhere after the operator cleared the per-customer floors. Register 134 -> 134 (opened 4, closed 4).
Report: REPORT-power-cut-fixes-2026-10-14.md. Both plugs ON.
Claude-Session: https://claude.ai/code/session_012qRErfCoiTkvDK9N5XHbzb
The felhom-build-deploy skill still showed the pre-R-840 copy-and-restart agent install; it now
leads with release-agent.sh + felhom-opsign agent_update / agent_config_update, as measured on
three boxes 2026-10-13.
03 §6 lists /guest/reboot and /host/poweroff; 13 counts 38 local-API routes and names the new
power of attacker (e). Hub CHANGELOG corrected: os_ring0_stalled is NOT held (a fleet rollout
alarm, on purpose). R-983: the GET reboot fixed (six remain). R-1000: the website FAQ line.
noindex and the robots Disallow lines removed; both URLs in the sitemap with their hreflang pair; the pair
moves from UNLISTED to TWINS in site_gates.py. The explaining sentence's base figure was already the drawn
box's base; a test now pins it (red-proved).
Empty or tiny selections drew a ~2.3 GB box whose base system (1,636 MB) no real box
has; the bar is now floored at the smallest named size (8 GB) for its parts only — the
recommendation and the rule are unchanged. Grouping applies only to two or more
leftover apps; a ninth app is shown by name in the group's colour. Tests: 12, the
new ones red-proofed.
The bar keeps its length (the fits-well box memory) and is split into base system
(box - guest), Felhom's own services, each app's peak (largest first, the dashboard's
palette minus amber and its first blue; beyond 8 apps one group) and spare room
(guest - peak) — all from the one rule; a test proves the parts sum to the bar within
1 MB. A legend under it (not sticky) and one sentence explain the recommended size.
Typical figure moved into the method fold-out. English: 'an 8 GB machine'.
Each category is a native <details>, closed at first, with an 'n selected' count;
a preset opens what it ticks. A sticky memory bar (0-64 GB, marks 8/16/24/32/48/64+):
solid fill = smallest box memory where the selection fits well, thin mark = where it
fits, but tight; one line names the box to buy. minBoxGb() inverts guestMib() by
bisection on it, so the rule is still written once. Box sizes 12/24/48/64 added;
tests cover seven sizes and the inverse (red-proofed); SIZING.md table gains 24 GB.
A household ticks apps and sees, for an 8, 16 and 32 GB box: fits well / fits, but
tight / does not fit, the totals, and the apps that burst on CPU. One rule in
assets/sizing.js, every number from assets/sizing-data.json (built by
scripts/build_sizing_web.py from the 2026-10-13 N100 measurement; test: 7 cases,
red-proofed). Unlisted: noindex, robots Disallow, not in the sitemap, linked from
nowhere — site_gates.py gate 25 enforces it (UNLISTED twin pairs keep every twin
check). SIZING.md §5 corrected: the 8 GB line contradicted its own x1.5 rule; guest
sizes 4.0/11.4/27.4 GB with the measured 612 MiB MemTotal shortfall. R-993 gains the
fact that the installer accepts a host below any minimum.
Measured on demo-felhom (N100, 16 GB): 61 apps deployed and removed through the
product, sampled every 5 s; heavy apps with one real action (Immich 200 photos,
Jellyfin transcode, Paperless OCR, Nextcloud 500 MB, ...). sizing.json for the
estimator (tools/build_sizing.py), SIZING.md with the proposed minimum machine,
METHOD.md. Power is RAPL CPU-package only (no meter). R-993 asks the operator to
approve the minimum machine; R-994 lists catalog limits/requests that are too low;
R-995 records leftover unnamed volumes after removal. R-231's dated check measured:
dooplex-backup.service success, journal green.
The catalog published SABnzbd and NZBGet today (app-catalog 8ee687e). Here:
- two cards in the media category on both language sets, each with the project's own logo and
each saying IN BOLD that a paid Usenet subscription is the household's own
- the published count 60 -> 62 everywhere it is stated; the category badges are derived from
the cards now (media read 15 with 17 cards, otthon/home read 14 with 16 — both stale)
- documentation/audits/usenet-2026-10-12/: both bench runs, both box steps and verdicts, the
nine defects with their measurements, what a missing provider left unmeasured, the teardown
- R-988 closed (moved to CLOSED-ITEMS in this commit); R-989 filed: the drill-catalog route of
09 §6.5 cannot be followed — that repo is private and no box has catalog credentials, so the
templates were staged in the box's own catalog cache instead
- walk.app_env: a SECRET field in app.yaml is stored ENC:, so that helper returns ciphertext —
the docstring says so now, measured the hard way
- STATUS.md and REPORT-usenet-2026-10-12.md
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117kRTSgs3rJWUFGxyoWN6Y
Read-only. fetch.py (polite, 5 s apart, 34 pages, all 200), parse.py (own text only; four parser
defects found by the controls and fixed), build.py + spotcheck.py (stdlib). tags.csv holds post
numbers and tags only - no username, author id, salt or post text in git. FINDINGS.md (2,908 posts,
195 people, 82 beginner threads in 7 clusters), FELHOM.md (catalog gap, hardware fit, 15 website
questions, objections, ten ranked recommendations), METHOD.md (controls, spot-check: role and
hardware below 85 % after two rounds, the table-driving splits 92.5 % and 97.3 %), REPORT.md.
R-990: operator rules on the recommendations. R-991: no buyer-readable hardware floor.
Claude-Session: https://claude.ai/code/session_012qRErfCoiTkvDK9N5XHbzb
Read-only. fetch.py + tables.py (stdlib), TABLE.md (45 charts, % of
respondents), FELHOM.md (apps, features, reasons, hardware, trends, ten
ranked recommendations), METHOD.md (hashes, counts, matching, MIT licence,
limits). Survey files are not committed. One register row (R-989) and one
STATUS line ask the operator to rule on the recommendations.
The catalog published qBittorrent and Transmission today (app-catalog 9081a47). Here:
- two cards in the media category on both language sets, each with the project's own logo,
each saying what a closed peer port costs (slower swarms, no incoming peers)
- the published count 58 -> 60 on every page that states it, FAQ JSON-LD with its visible twin
- documentation/audits/qbittorrent-2026-10-12/: both bench runs, both box walks and verdicts,
verify-fixes.json, teardown-newapps.json, and the README brought to its final state
- the 2026-09-30 walk harness: a `type: path` deploy field is filled with the drive ROOT —
the controller refuses a path inside a drive (409), so the walker could not install any app
with a drive field at all
- R-988: the Usenet pair (SABnzbd, NZBGet), operator request, not started
- STATUS.md and REPORT-newapps-2026-10-12.md
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0117kRTSgs3rJWUFGxyoWN6Y
- 09 §3: the decision sheet S1-S38, the operator's words quoted where they differ (S1, S2, S8, S32, S37)
- closed: R-831 R-870 R-698 R-327 R-230 R-719 R-352 R-526 R-770 R-771 R-913 R-290 R-209a R-769 R-616 R-884, and R-243 (dated check: the mail fired 2026-10-11 16:13:51Z, hub log agrees)
- R-698: the known limit written in 07 §6.6; new row R-933 (mirror supported catalog images, P3)
- R-327: capability-map row BUILT; stands claim re-titled and moved to built; page re-rendered
- R-913: the customer Cloudflare token recipe in day0-install A.1; 01 §7 points at it
- R-884: read only - ArgoCD Synced, live Deployment equals git
- R-887 dated check NOT met (4 lost jobs 10-10/11); re-dated 2026-10-19 after the S21 change
- 22 open rows carry their ruling; STATUS: the sheet is ANSWERED
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
documentation/audits/mockups-2026-10-11/
COMPARE.md the three directions side by side, measured the same way:
A current preview 12 req / 270 KB
M1 dashboard 68 req / 1012 KB (front page 1 660 px)
M2 hybrid 16 req / 150 KB (phone 3 057 px, the shortest)
plus three sentences each (good / risk / cost) and the risk I
think is real: a non-technical visitor may read M1 as somebody
else's account and look for a login.
SHARED-DESIGN.md REPORT ONLY, nothing built. M1 needed 260 CSS rules: 188 (72%)
copied from the controller UNCHANGED, 8 adapted, 64 new. It used
188 of the controller's 826 rules (23%). The 18 tokens are
already identical - verified by parsing both :root blocks, zero
differing values.
Recommends (a) a canonical felhom-ui.css plus a copy-drift gate,
over (b) serving the controller's stylesheet to the website -
and recommends NOT building it until a direction is chosen.
shots/ desktop and phone captures of all three, from the deployed pages
The finding that outranks the percentage, and the reason for the recommendation: a
component-level copy silently drops the element-level rules those components depend
on. The extractor copied every .sidebar/.content/.mobile-topbar rule and missed
"@media(max-width:768px){ body { flex-direction: column } }". Everything LOOKED
copied; the phone layout was broken in four measurable ways. Invisible to a diff,
invisible to review. A second instance the same evening: the drawer is gated on a
'js' class set by an inline script that was not copied either.
Register 129 -> 129 (one closed, one opened):
R-929 CLOSED - NOT A DEFECT. Measured from the catalog: 60 templates, minus
plant-it (abandoned) and wger (hidden) = 58 offered, exactly what the prose
says; the page's 59th card is FileBrowser carrying the single "Beepitett"
badge. Set difference both ways is empty. The premise was wrong on BOTH
halves. Rule kept: before filing a count disagreement, check whether the
counts are counting the same set.
R-932 OPEN - 8 catalog strings across 5 apps carry an em dash, and they render in
the controller's app page on every real box. Found because the mockups are
the first thing to show catalog copy on a website page. Lives in a repo this
session did not have in scope, so it is a row, not a fix.
PREFLIGHT.md - taste-skill §14 plus the Felhom additions, run in writing against
the DEPLOYED preview, Pass/Fail per line. Measured, not assumed:
page height 1440: 10 922 -> 6 780 px (38% shorter)
page height 390: 19 913 -> 10 865 px (45%, 23.6 -> 12.9 phone screens)
initial load: 13 req / 308 KB -> 11 req / 260 KB (same conditions)
fully loaded: 16 req / ~540 KB -> 15 req / ~545 KB (ESSENTIALLY UNCHANGED -
the 93 KB portrait is paid for by dropping the 49 KB hero logo
and the 31 KB mono font; stated rather than spun)
third-party hosts: 0 -> 0
em dashes in our copy: 18 -> 0
layout families: 8 for 11 sections (3 repeats) -> 10 for 10 (none)
no horizontal scroll: clientWidth == scrollWidth at 390 and at 1440
Contrast: three NEW styles measured 3.88 / 3.14 / 3.57 on the first pass, all under
WCAG AA. Fixed to --text-2 and re-measured on the deployed file: 7.56 / 6.12 / 6.96.
A measurement that lied, recorded because it nearly shipped a false verdict: the
first helper reported .limits p at 1.17:1 by treating the translucent --warn-dim
as opaque. Compositing the whole ancestor stack gives the true 5.67:1, a pass. The
three real failures in the same run were real. An instrument that can report a
wrong number silently is not a measurement.
Two rows filed, both needing something this session may not do alone:
R-930 (Business & legal, P3) the primary CTA is 4.01:1, under AA. MEASURED
IDENTICAL ON THE LIVE PAGE, so pre-existing and site-wide, not a regression
of this redesign. Three fixes offered, each touching brand, so it goes to
the operator rather than being changed unilaterally.
R-931 (Process & tooling, P4) felhom.eu serves NO compression at all - site.css
goes over the wire as 94 310 raw bytes with no Content-Encoding even when
gzip is offered. An nginx setting in manifests/, not a website change.
Worth more than anything on this page.
Register: 127 -> 129 rows, 2 opened, 0 closed. register_shape_gate.py OK.
OUTREACH.md: a personal message, a Facebook group post with a shorter variant,
a forum post, and a table of 15 places with the rules quoted where they are
published. Nothing was posted and no group was joined; the group data was read
from each group's public About page on 2026-10-10.
The rules that matter, measured rather than assumed: Magyar Linux Felhasznalok
(11970, public) has no advertising ban and Felhom is on topic; DIY Smart Home
forbids ads and names an e-mail route; HUP forbids unlicensed advertising; the
prohardver family forbids "tagok, vasarlok gyujtese ... felhivas" without the
operator's permission, which is literally what recruiting testers is.
R-929 (P4): the website's prose says 58 apps, the apps page renders 59 cards,
the catalog holds 60 template directories. Gate 15 cannot catch it because it
compares the two language sets to each other, never to the catalog. Not fixed
here - it needs the catalog skill's exclusion rules. The campaign posts say 58,
matching every prose claim on the site.
1631 (catalog b7f0f7c), 1632 (felhom.eu 970616b7) and 1634 (felhom.eu 29c9cb1e):
every step failure with 0 s after 10.5-14 minutes in set-up, logs absent. A job
never picked up would end at an arbitrary second; the same second-of-the-hour
twice two hours apart points at a periodic sweep or a fixed deadline. Between
them the runner worked — catalog job 1633 succeeded normally on the same tree,
and a shallow clone with no sibling runs the catalog gates green by hand, so the
code is not what is red. The 2026-10-12 due-check's premise is false and the row
now says so.
The operator was asked at the end, as the brief said, and ruled: keep Monica
out. R-927 opened and closed in the same session, with the measurements and the
sentence worth keeping — an app is not added because a gap exists, it is added
because the app can be kept. Register 137 -> 138: 2 opened, 1 closed.
Also in here: the two small things seen and not filed (remove leaves hold-logs
in the stack directory, with no secret in it; the catalog clone's pre-push hook
is unarmed, so the gates were run by hand before the commit and by CI after).
The catalogue side is app-catalog-felhom.eu b7f0f7c. Here: the evidence, the
website, the register and the operator's view.
Website
- Two cards in the Otthon & Eletmod / Home & Lifestyle section of BOTH apps
pages, with assets: grocy-logo.svg is grocy's own icon with its single fill
made white like the other logos, lubelogger-logo.png is the app's own icon
with its dark background dropped and the mark made white (the rule
SparkyFitness's PNG follows), and six screenshots of each app's own UI with a
household's own data, taken headless on the bench from the published template.
- The app count moved 56 -> 58 in 15 places per language set, both languages,
and the open-source tile 49 -> 51. Checked by asking the same patterns for the
new number afterwards. marketing/facebook/COPY.md still says 56 and is NOT
changed: the post it carries is already scheduled.
Evidence
- documentation/audits/new-apps-2026-10-10/ — FIT.md (checklist group 0 for all
three, with the Hungarian-UI column), the bench and box transcripts, the
memory samples, the screenshots and the gate runs.
Register: 137 -> 139 rows, 2 opened, 0 closed.
- R-926 after a remove-keeping-backups and restore, nobody has checked what the
app page shows for an after_install app's generated password. Measured here:
LubeLogger is safe (its login is derived from the environment at every start,
the new password signs in, the data is back); grocy's install password still
signs in from the restored database but the deployed environment no longer
carries ADMIN_PASSWORD at all. Seven apps are in the class.
- R-927 Monica, stopped at checklist 0.2 with the measurements, waiting on the
operator.
Gate scripts: the same console trap in nineteen of them and in repo_gates.py
itself, where it ABORTED THE WHOLE RUNNER at the first gate — printing a
non-ASCII character on this workstation's cp1250 console raised
UnicodeEncodeError before the gate had decided anything, and reuse_refs_check.py
died while printing a NOTE. All now reconfigure their own streams. Red-proof
that the remaining script-test failures are not mine: test_due_checks_gate.py
fails the same 5 of 42 with the change reverted.
admin
d