The full pre-release smoke still expected #198's T-node fixture to compact into one record. #299 intentionally preserves the outer/shared ownership breakpoints, so the smoke now requires three canonical 22 cm role runs while retaining Preview, atomic Apply, Reload, and Undo coverage.
Issue: #299
User-Visible: no
Canonical Linux artifact from Validate run 32775157799 was captured from the final combined dev SHA 8a3a115. All seven material changes were visually inspected: two hidden-wall diagnostic scenes expose axes and nodes, two resize scenes add the approved measurements and area labels, and three wall drawing/junction scenes expose the approved diagnostics. The complete 110-scenario artifact is accepted without partial replacement.
Issue: #296
Issue: #300
User-Visible: no
Release: v1.67.0-rc.1
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32775157799
Canonical Linux capture from workflow run 32774826636 checked out exact branch SHA 5696e274fe. All ten frames were reviewed against the current accepted set: nine are byte-identical, while 06-device-editor differs by one antialiasing pixel and remains visually unchanged.
Issue: #299
User-Visible: no
Owner-approved review exception: the external reviewer is unavailable. The exact branch SHA passed typecheck, 1287 unit tests, bundle parity, targeted browser smokes, six 24-step edit walks, late traces, mutation testing, performance comparison, and the canonical Linux documentation capture. The merge also removes one trailing blank line from the specification; product sources are unchanged from the validated branch.
Issue: #299
User-Visible: no
The full smoke suite still expected Escape to undo one wall point. The accepted #294 contract finishes and detaches the chain while Ctrl/Cmd+Z owns undo; the dedicated wall-tool smoke already proves geometry persistence.
Issue: #294
User-Visible: no
Canonical Linux capture from workflow run 32772784765 was visually inspected after integrating #296, #298, and #300. It refreshes the exact combined source fingerprint and all ten accepted frames.
Issue: #300
User-Visible: no
Owner-approved review exception: the external reviewer is unavailable. The exact branch SHA passed its recorded gates; the integration keeps #296 diagnostic geometry above masonry and #300 resize measurements above wall bodies. Generated bundles were rebuilt from the combined sources.
Issue: #300
User-Visible: no
Owner-approved review exception: the external reviewer is unavailable. Both r1 High findings were fixed, the exact branch SHA passed all recorded gates, and generated bundles were rebuilt after conflict resolution with #296.
Issue: #298
User-Visible: no
Owner-approved review exception: the external reviewer is unavailable. The exact branch SHA passed its implementation gates and the issue records the evidence.
Issue: #296
User-Visible: no
Смок импортировал `../test-build/*` статически и на моей стороне работал только
потому, что каталог остался от `npm test`. На чистом Linux CI его нет: в job
`smoke` идут `npm ci` и `npm run bundle:sync`, сборки тестов там не бывает, и
смок падал на импорте до запуска браузера.
Ровно эту ошибку уже проходил `smoke_lattice_write_barrier.mjs` — там об этом и
написано в комментарии. Лечение то же: собрать `tsconfig.test.json` перед
динамическим импортом.
Проверено в чистом worktree без `test-build`: шесть прогонов, все совпадают с
таблицей KNOWN, OK.
Issue: #297
User-Visible: no
Все прежние гейты проверяют снимок модели. Дефекты геометрии рождаются в
редактировании: #289, #296 и #298 прошли решётку, кладку, роли, ключи и аудит
ручек, потому что такая геометрия снимок не портит — она портит следующий жест.
demo/smoke_edit_walk.mjs расшатывает реальный план продуктовыми жестами
(_rszEdgeDown/_rszMove/_rszUp, _confirmRoomDelete, optimizePlans) по фиксированному
семени и после каждого шага судит конфиг в node. Второго представления редактора
не появляется — принцип #292.
Подшаговый шум остаётся наблюдением, а не нарушением: координата пишется девятью
знаками, 304/240 = 1.266666667, отклонение 8e-8 шага неустранимо форматом
хранения и уйдёт на этапе 1 ADR #282. Судится только «вне сетки».
Таблица KNOWN работает в обе стороны: обход падает и когда находок больше, и
когда меньше. Молча позеленевший гейт не сообщает о починке — так
partition-mt2on9ou-0 прожил в плане владельца от беты 9 до rc.1.
Новый инвариант checkHiddenObstacles: перегородка на стене комнаты, незакрытый
контур на стене комнаты, черновик, который не может стать комнатой (#296).
Найдено сразу, в пяти прогонах из шести — на первом жесте: #298 (ресайз уводит
конец записи толщины мимо решётки и мимо ребра), #299 («Оптимизировать» и
удаление комнаты сливают записи через границу роли).
Issue: #297
User-Visible: no
Keep unrelated pre-existing near-axis edges from disabling exact Resize handles, make tracked single-space fixtures visible to the invariants CLI, and add the three missing mutation gates plus real-plan coverage.
Issue: #290
User-Visible: no
Accepted the complete Linux artifact after visual review. Only the dark and light safe-resize handle views change intentionally; all other passing scenarios keep their prior bytes.
Issue: #289
User-Visible: no
Release: v1.67.0-beta.10
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32726613816
Accepted the complete Linux artifact after visual review. Only the dark and light isometric geometry views change intentionally; 106 passing scenarios keep their prior bytes.
Issue: #260
User-Visible: no
Release: v1.67.0-beta.10
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32725286757
Refresh the shared wall-key fixture contract and keep generated frontend bundles synchronized with the fingerprinted geometry fixtures.
Issue: #260
User-Visible: no
The continuity gate cannot see the defect from the owner's 66.json: masonry is
continuous there and the record agrees with what is painted. The record itself
is wrong — a partial resize left 43 steps of a former shared boundary as an
exterior wall while it kept the 20 cm of that boundary, next to 30 cm exterior
neighbours. Thickness followed the key, not the role of the edge.
A width check would not have caught it, and I built one before throwing it away.
It compares the painted body against the record, and here the two agree. On real
plans it also fires where masonry is legitimately wider — columns, junction
influence, abutting parallel walls: 76 to 82 steps measured against 4 expected,
every case legal. A gate that needs explaining half the time is noise.
The defect is expressible in a single state instead: one record whose span is
partly shared and partly exterior. Nobody sets that on purpose. Roles come from
the polygons — a stretch is shared when another room's edge covers it — so the
check needs neither a build nor product code.
Two traps found by measurement, both of which produced false positives. Count
distinct rooms rather than edges: in a corner one room owns two edges, and
counting edges called every exterior corner a shared boundary — 12 and 14 false
positives. And do not sample the endpoints: an endpoint is a node, where a wall
legitimately touches two rooms, and including them reported 95 per cent exterior
on every wall abutting a shared one.
Mutation coverage, stated honestly: the endpoint mutant is killed by the
real-plan test. The rooms-versus-edges mutant survived — once endpoints are
excluded, counting edges gives the same answer on real plans, so that choice is
not load-bearing. I removed the mutant rather than ship a surviving one, and said
so in the code.
Issue: #287
User-Visible: no
The second floor covers one class: the multi-wall corridor eating a neighbouring
wall. The first floor brings what neither it nor any synthetic fixture has —
three virtual spans, two wall columns, two solid edges with no thickness record
at all, and 127 noisy coordinates.
The two plans pin opposite states, which is worth more than two plans with the
same defect: one records a known debt of 181 steps, the other records cleanliness
at zero. A regression is caught in both directions.
A declared virtual span is a declared break, not a defect, so the rule from #285
would have reddened on the first floor's own contract. Samples lying on
open_spans are now excluded: 830 of 6800 on that plan, and the remaining gap
count is zero.
The two zero-thickness solid edges are two grid steps long and covered by the
bodies of their neighbours, so the browser sees no break. Their count is
therefore pinned in the model test rather than the smoke — if it grows, or if the
neighbours stop covering them, that shows up before it becomes a hole.
Both fixtures must stay noisy: the project's synthetic models carry exactly zero
noise, which is why #258 and #248 are not reproducible on them. A profile check
asserts at least a hundred noisy coordinates each, so a future write barrier run
over the fixtures cannot quietly rob them of their purpose.
Privacy: names, ids, markers, device bindings and layout are gone; geometry
stays, coordinates unchanged, because they are the point.
Issue: #286
User-Visible: no
Eight closed issues on wall junctions — #271, #272, #275, #276, #277, #278,
\#279, #280 — shipped in beta.9, and the break in the owner's real plan
survived. Every one of them was accepted on synthetic fixtures: a cell-5
mixed-depth T, a rectilinear T with three equal half-depths. On those the fixes
work.
The smoke asks the product itself, through isPointInFill on the wall path,
whether masonry exists where the model promises it. That is independent of both
resolution and the pixel-diff thresholds which miss this class: a 45-step break
on a large plan is a fraction of a per cent of the frame, well under the 0.05
per cent scene tolerance.
Measured on the shipped code: four breaks, 181 grid steps in total, 45.25 each.
That number is derivable rather than incidental — the node joins a 30 cm
exterior wall, a 30 cm spur and an arm five steps long; half-depth 15,
MITRE_LIMIT 4, corridor radius 60, and 60 − 15 = 45 steps are cut out of the
neighbouring 20 cm wall. The corridor eats the masonry, which is what #271 and
\#275 describe.
The fixture is privacy-minimised — neutral room names and ids, no device
bindings — and keeps its coordinates, because they are the point. It lives in
test/fixtures rather than demo/fixtures on purpose: sourceFingerprint hashes the
.mjs of demo/fixtures and demo/golden, so anything added there staleizes the
committed bundle and the screenshot manifest at once. Verified after this
commit: bundle still fresh, screenshots still current.
The debt is recorded as numbers and compared exactly. Better and the test asks
for the numbers to be updated, which proves the improvement; worse and it
catches the regression. Verified by execution in both directions.
Issue: #285
User-Visible: no
Stage 0 of ADR #282. A lattice node is k/240, which has no exact binary
representation, and a stored coordinate is a float. Nobody could say how much of
a real plan is affected, and Optimize promises to remove coordinate noise
without a definition of noise that can be checked.
latticeProfile splits every coordinate of the model into three populations,
because they are three different problems: exactly on a node, near a node but
not exact, and legitimately off grid. The middle one is the defect class behind
\#258, \#279 and the non-converging Optimize; the last one is authored geometry
the current model allows and must not be called a violation.
Measured on the owner's installation: space 1 has 208 coordinates, 33.65 per
cent exactly on a node and 65.38 per cent in the noise class; space 2 has 21.23
against 78.77. The worst deviation is 8e-8 of a step — invisible, and enough to
put a wall key in the neighbouring bucket.
The counterpart is what makes it worth having: every shipped fixture has zero
noise, all of its off-grid values being authored. Our own test data therefore
cannot reproduce this class by construction, which is why the owner finds these
defects and the gates do not. A test pins that property so it cannot drift.
No violations are produced, no gate turns red, and nothing is repaired: what to
do with a vertex 8e-8 from a node is the owner's decision, and this measures its
price first.
Issue: #283
User-Visible: no
Nine of the last ten specs are one class of defect, 23 fix commits in thirty
days, 595 tolerance mentions across ten geometry modules, one every seventh line
in resize.ts. The specs rate their own risk at 9 and 10 of 10, and two titles
show where that arrived: a nearly orthogonal T junction, and a union failure
that must merely damage less.
The ADR names three properties of the data model that produce the stream — a
float coordinate on a 1/240 step, an identity derived instead of stored, and a
wall living in two representations at once — and records four stages against
them. Stage 0 is accepted for implementation; the rest are direction, in the
same sense as #34.
Issue: #282
User-Visible: no
Package the reviewed Optimize reconciliation, safe fixed-topology Resize and wall-union isolation fixes from #276, #277 and #278 as the eighth v1.67 prerelease candidate.
Issue: #278
User-Visible: yes
Accept the complete 106-frame Linux golden artifact. All 104 existing scenarios passed; six byte-only PNG refreshes have zero pixel diff, and both new safe-resize light/dark candidates were manually reviewed.
Issue: #277
User-Visible: no
Release: v1.67.0-beta.8
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32689229827
Use paired ABBA/BAAB batches for the p95 gate and accept the complete 104-frame Linux golden artifact. The four new #276 scenes were visually reviewed; every existing scene passed, and the nine byte changes contain zero pixels above their comparison threshold.
Issue: #276
User-Visible: no
Release: v1.67.0-beta.8
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32684802336
Accept the complete ten-frame Linux artifact captured from candidate 8d2a050. All five changed frames were reviewed against the committed set; they preserve the intended UI while recording the tapered #272 geometry and beta.6 source fingerprint. Run: https://github.com/Matysh/houseplan-card/actions/runs/32667107212
Issue: #272
User-Visible: no
Keep the finite #272 exit as a safe subset of the square connector while removing two contour corners per bevel. The simpler canonical path preserves the zero-hole and finite-ray contracts and restores headroom in the large-house Glow gate.
Issue: #272
User-Visible: no
Accept the complete ten-frame Linux artifact from the exact #272 performance gate-fix source. The content matches the reviewed UI while replacing the non-canonical Windows captures that had entered dev with #274.
Issue: #272
User-Visible: no
Subtract each node's combined local bevel mask once instead of traversing the large wall and paper geometry once per triangle and exterior connector. The equivalent mask keeps the reviewed #272 geometry while restoring the large-house Glow state-update budget.
Issue: #272
User-Visible: no
Accepted the complete 98-scenario Linux artifact after visual review. Intentional differences are limited to the finite degree-3 junction repair and the bounded multi-wall bevel closures; all remaining candidate bytes come from the same complete canonical capture.
Issue: #271
Issue: #272
User-Visible: no
Release: v1.67.0-beta.6
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32661228757
Accepted the complete 98-scenario Linux artifact after visual review. The intentional changes are the new wall-key round-trip regression scene and four large-house frames where #258 restores the fixture's real walls; all 93 passing candidates remain on their prior bytes.
Issue: #258
User-Visible: no
Release: v1.67.0-beta.5
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32649309598
git grep _bindingCandidates -- test/ demo/ was empty: the function that decides
what the user is offered under Add had no test and no smoke. Tombstones were
covered from every side, the list they filter was never asked. That is how #262
reached us through a user report instead of a gate — smoke_hidden_flag assigns
binding straight into _markerDialog and bypasses the picker entirely.
Twelve checks against the real bundle: a deleted device is offered again, a
deleted plain entity is offered again behind the checkbox, the checkbox itself
is the trap (a device entity is absent with it off, present with it on, and it
starts off for a new marker), a placed binding is not duplicated, and re-adding
replaces the tombstone and leaves the picker.
The twelfth pins the known defect #262 as current behaviour: a device tombstone
still hides its child entities. Fixing it turns the check red and forces it to
be flipped, so the fix cannot pass the coverage by.
smoke-links registers only the pure tombstone helpers. The picker names itself
and is found by direct match; the helpers are not named anywhere in the
scenario. Verified by probe: touching src/devices.ts alone selects this smoke
as a registered link, and without the entry nothing would select it.
Issue: #263
User-Visible: no
The first cut of checkWallKeys compared the stored key against endpoints
snapped to the lattice and called any mismatch a violation. Both halves were
wrong, and measurement says so: wallIntervals reports the query key for the
disputed edge as 0.887500,0.195833@1.5706, i.e. the form built from the
coordinates as stored, and the two owner configurations that differ in exactly
these keys produce byte-identical wall bodies and multi-wall node maps. The
check would have reddened a plan that renders correctly.
Graded now: a drift inside the tolerant fallback's half-pitch reach is an
observation, a key beyond it or one that does not parse as coordinates is a
violation. The threshold is expressed in grid steps with a 1e-3 slack — with a
relative 1e-6 the four identically drifted records of one plan split between
the two classes on their last bits, so the check repeated the very rounding tie
it exists to expose.
visual-matrix leaves KEY_CONTRACT_DEBT: its keys drift inside the reach and now
read as observations. large-house stays — its labels do not parse, and
wallIntervals shows all 80 solid edges resolving to zero thickness (#260).
Issue: #259
User-Visible: no
#258 lost two thickness records to a rounding tie: wallKey quantises the
midpoint with Math.round, and a wall of odd step length has its midpoint
exactly on the tie, where the exact node 83/240 and the stored 0.345833333
fall on opposite sides. The #254 invariants pass on that file — their edge
tolerance is 0.004 and the drift is 0.00417, so the check sits on its own
boundary.
checkWallKeys compares strings, and against endpoints snapped to the lattice:
keying from the raw endpoints flags the healthy state and passes the broken
one, which is what the first formulation in #258 got wrong. Measured on the
owner's before/after pair: 0 findings before, exactly the two artefact walls
after.
Two shipped fixtures write keys off the contract (#260); recorded as a number,
so the debt can neither grow nor be silently fixed.
Issue: #259
User-Visible: no
Accepted the complete 97-scenario Linux artifact after visual review. The intentional golden changes are limited to the EN/Dark and RU/Light Optimize orphan-cleanup dialogs; all 95 passing raster candidates were restored to their prior bytes and hashes. The canonical docs artifact used the same Chromium and refreshed its source fingerprint; eight frames were identical, while two differed by only 2 and 17 sub-threshold pixels.
Issue: #252
User-Visible: no
Release: v1.67.0-beta.4
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32623704126
Ревью шло по ветке как есть, слияние делало ребейз: проверенный SHA и
слитый SHA были разными коммитами. Текстовое расхождение ловил конфликт,
смысловое git склеивал молча — так пришёл регресс #234. Заодно конфликт
обнаруживался после сорока минут работы ревьюера, хотя виден до них.
Новый шаг для этапа code, сразу после выбора ветки: потомок dev —
ничего; отстала и ребейзится — ребейз, push с --force-with-lease, ревью
приведённого состояния и запись о ребейзе в промпт (§7.2 требует полного
разбора); конфликт — возврат в S6-in-progress без запуска ревью.
Issue: #257
User-Visible: no
(cherry picked from commit 793a6486d8)
Update the documentation capture source fingerprint after the golden-matrix correction. All ten canonical frames were reviewed; two sub-threshold raster-noise candidates were kept byte-identical to HEAD and only the manifest changed.
Capture: https://github.com/Matysh/houseplan-card/actions/runs/32621056471
Issue: #251
User-Visible: no
Accept the complete 96-scenario Linux artifact after the #251 rebase exposed the unaccepted #249 junction visuals. Reviewed changes are limited to thirteen bounded-junction frames and the new multi-wall bevel scene; all passed raster-noise candidates were restored.
Issue: #251
User-Visible: no
Release: v1.67.0-beta.4
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32620456718
Use the discarded-wedge centre derived by the #249 unit geometry instead of a point that lies inside the legitimate incident wall body. This restores the semantic golden contract exposed while rebasing #251; product rendering is unchanged.
Issue: #251
User-Visible: no
Keep the cover-precedence scenario focused on target state mirroring by explicitly providing a live own entity after #251 separated controller availability from target availability.
Issue: #251
User-Visible: no
Accepted the complete Linux artifact after visual review. Only device-icon-state-table light/dark change intentionally; seven within-threshold renderer-noise images were restored to their prior bytes and hashes.
Issue: #251
User-Visible: no
Release: v1.67.0-beta.4
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32617372743
Close SPEC-REVIEW-244-r1 M1 by making restored-marker parity in desktop, touch, kiosk, and Static explicit while retaining the desktop-first editor contract.
Issue: #244
User-Visible: no
Specify deterministic Optimize and import repair, safe marker detachment, guarded space deletion, and invalid default-floor feedback for issue #244.
Issue: #244
User-Visible: no
Accepted the complete Linux artifact after visual review. The intended visual changes are the decor layer above room and Glow-base fills (#231), centered opening symbols with preserved flip direction (#242), and exact before/after space-tab drop indicators (#243). The unchanged large-house zoom 0.40 frame remained within its existing threshold and was restored to its prior bytes and hash.
Issue: #231
Issue: #242
Issue: #243
User-Visible: no
Release: v1.67.0-beta.2
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32597653292
Accept the canonical Docs screenshots artifact after visually reviewing the opening-symbol geometry in the changed plan editor frame.
Issue: #242
User-Visible: no
Address code review H1 by keeping gate flip direction observable without a second vertical mirror. Add fail-closed golden contracts, smoke coverage, and a mutation guard for the affected opening-symbol geometry.
Issue: #242
User-Visible: yes
Accepted the complete Linux artifact after visual review. The intentional changes are limited to Optimize preflight failure dialogs, opening inner-distance overlays, and the corrected live wall-thickness preview. Renderer-noise images that stayed within their existing thresholds were restored to their prior bytes and hashes.
Issue: #199
Issue: #234
Issue: #238
User-Visible: no
Release: v1.67.0-beta.1
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32576969813
Spec review r1 returned two blocking findings and both were right.
A measured side that is itself a passage would have been shortened by its
neighbouring walls, while insetContour — the very function the area label
already uses — treats that joint as a flat cap and shortens nothing. Length
and area would have diverged again, at a different boundary, which is the
defect this task exists to remove. The zero rule now comes first and returns
the full centreline length for an open side.
The thickness source was wrong as a matter of fact, not of taste: an
existing test shows thicknessCmAt returns 0 for a whole-edge query against a
partially set thickness, so a split-thickness edge would have silently
stopped shortening. Half-depths now come from roomWallProfile, the atomic
profile that innerContourForRoom already uses for the area, so one edge is
resolved by one mechanism.
Two acceptance criteria and two mutation guards added for the closed
findings.
Issue: #233
User-Visible: no
The contract named benchmark and golden tooling, which is exactly why the smoke
launcher was allowed to skip the check for so long. It now covers every browser
check, names where each one gets it, and records why a stale bundle is worse
than a plain failure: part of the assertions go red and part stay green.
Issue: #236
User-Visible: no
Golden runs, benchmarks and documentation captures each called
assertFreshDemoBundle; the smoke launcher never did, so all ~128 smokes could
silently test a stale demo/srv/assets bundle. On #234 that cost a round of
analysis: three assertions went red and a fourth went green, because the old
code was wrong in two places that agreed with each other, and a mixed result
reads as a logic defect rather than a stale artefact.
launch() now runs the check once for every smoke, against the repository root
rather than the serving root — demo/srv has no src/** to fingerprint.
HP_ALLOW_STALE_BUNDLE=1 skips it for debugging and warns out loud, because a
guard that says nothing when it steps aside is the silent success this project
keeps removing. A mutation entry proves the call cannot quietly disappear.
Issue: #236
User-Visible: no
Golden runs, benchmarks and documentation captures each called
assertFreshDemoBundle; the smoke launcher never did, so all ~128 smokes could
silently test a stale demo/srv/assets bundle. On #234 that cost a round of
analysis: three assertions went red and a fourth went green, because the old
code was wrong in two places that agreed with each other, and a mixed result
reads as a logic defect rather than a stale artefact.
launch() now runs the check once for every smoke, against the repository root
rather than the serving root — demo/srv has no src/** to fingerprint.
HP_ALLOW_STALE_BUNDLE=1 skips it for debugging and warns out loud, because a
guard that says nothing when it steps aside is the silent success this project
keeps removing. A mutation entry proves the call cannot quietly disappear.
Issue: #236
User-Visible: no
Golden runs, benchmarks and documentation captures each called
assertFreshDemoBundle; the smoke launcher never did, so all ~128 smokes could
silently test a stale demo/srv/assets bundle. On #234 that cost a round of
analysis: three assertions went red and a fourth went green, because the old
code was wrong in two places that agreed with each other, and a mixed result
reads as a logic defect rather than a stale artefact.
launch() now runs the check once for every smoke, against the repository root
rather than the serving root — demo/srv has no src/** to fingerprint.
HP_ALLOW_STALE_BUNDLE=1 skips it for debugging and warns out loud, because a
guard that says nothing when it steps aside is the silent success this project
keeps removing. A mutation entry proves the call cannot quietly disappear.
Issue: #236
User-Visible: no
Golden runs, benchmarks and documentation captures each called
assertFreshDemoBundle; the smoke launcher never did, so all ~128 smokes could
silently test a stale demo/srv/assets bundle. On #234 that cost a round of
analysis: three assertions went red and a fourth went green, because the old
code was wrong in two places that agreed with each other, and a mixed result
reads as a logic defect rather than a stale artefact.
launch() now runs the check once for every smoke, against the repository root
rather than the serving root — demo/srv has no src/** to fingerprint.
HP_ALLOW_STALE_BUNDLE=1 skips it for debugging and warns out loud, because a
guard that says nothing when it steps aside is the silent success this project
keeps removing. A mutation entry proves the call cannot quietly disappear.
Issue: #236
User-Visible: no
Golden runs, benchmarks and documentation captures each called
assertFreshDemoBundle; the smoke launcher never did, so all ~128 smokes could
silently test a stale demo/srv/assets bundle. On #234 that cost a round of
analysis: three assertions went red and a fourth went green, because the old
code was wrong in two places that agreed with each other, and a mixed result
reads as a logic defect rather than a stale artefact.
launch() now runs the check once for every smoke, against the repository root
rather than the serving root — demo/srv has no src/** to fingerprint.
HP_ALLOW_STALE_BUNDLE=1 skips it for debugging and warns out loud, because a
guard that says nothing when it steps aside is the silent success this project
keeps removing. A mutation entry proves the call cannot quietly disappear.
Issue: #236
User-Visible: no
The spec review found the i18n section missing outright — the analysis
comment claimed it was untouched, the document itself said nothing, and a
DoR section cannot be inferred from a comment. It now says so explicitly,
and adds that the absence of i18n files from the diff is part of the
contract rather than an accident.
The waived Low is closed too: the old wallChainSegments treated a recorded
zero as valid while the new contract requires strictly positive values, so
zero is now named in the AC1 examples instead of being derivable from the
prose.
Issue: #234
User-Visible: no
Гейт мутаций работает в изолированном `git worktree`, где `test-build/` не
существует: юнит-гвард, идущий сразу в `node --test`, падает с
`ERR_MODULE_NOT_FOUND` ещё на чистом прогоне — то есть не проверяет ничего, а
еженедельный прогон реестра останавливается на первом же таком мутанте и не
доходит до остальных. Соседние юнит-мутанты этого не допускают: их гвард
начинается с `npx tsc -p tsconfig.test.json && node scripts/fix-test-build.mjs`.
Четыре мутанта #230 теперь тоже.
Проверено штатным харнессом, а не вручную: `node scripts/mutation-gate.mjs
--id=<каждый>` — «поймано 1 из 1» для всех семи мутантов задачи, включая три
смоковых, которые и раньше работали.
Тем же дефектом страдают юнит-мутанты #220 и #229 — это чужой скоуп и предмет
[#235](https://github.com/Matysh/houseplan-card/issues/235); здесь не трогаю.
Issue: #230
User-Visible: no
Две сцены и только они: `large-house-zoom-040-dark` (шаг штриховки был 20
юнитов, стал 8) и `large-house-zoom-250-dark` (был 3.2, стал 8). Расхождение
осмотрено покадрово: меняется только плотность штриховки тел стен, колонн и
перегородок — геометрия, цвета, устройства и свет идентичны. Это прямое
следствие решения владельца §4.2 ТЗ, ради которого зумовая компенсация и
убиралась.
Принято `npm run golden:accept -- --reviewed`. `accept` переснимает весь набор,
поэтому пять сцен, разошедшихся на шуме рендера
(`isometric-large-warm-remount-dark`, `room-label-parity-plan-dark`,
`tray-medium-group-en`, `wall-junctions-plan-preview-light`,
`wall-junctions-plan-t-dark`), возвращены к прежним байтам вместе с их хэшами
в индексе: `verify` считал их совпадающими в пределах допуска, и принимать их
задача #230 права не имеет.
Baseline-Reviewed — прогон, где job `golden` прошёл на Linux ровно на этих
эталонах.
Issue: #230
User-Visible: no
Release: v1.67.0-beta.1
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32480753934
`edf1cca` принял два эталона внутри продуктового коммита, а по правилу
процесса коммит, трогающий `demo/golden/baselines/**`, обязан нести `Release:`
и `Baseline-Reviewed:`. Локально это не остановилось: в моём клоне не был
выставлен `core.hooksPath`, и `commit-msg` попросту не запускался — теперь хуки
установлены, проверено повторным прогоном скрипта вручную.
История не переписывается (AGENTS.md: «never rewrite published history to
satisfy trailers»): эталоны снимаются этим коммитом и возвращаются следующим,
уже с положенными трейлерами.
Issue: #230
User-Visible: no
Release: v1.67.0-beta.1
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/32480753934
Одна и та же стена 15 см выглядела на планах с разным `cell_cm` по-разному:
шаг паттерна был константой в юнитах, а толщина стены переводится в юниты через
`cell_cm`, — значит число полос было пропорционально `1/cell_cm`. Разброс между
крайними масштабами достигал 25 раз, а при `cell_cm ≥ 10` в стену не попадало
и одной полосы: штриховка вырождалась в случайные штрихи или исчезала.
Шаг стал физической величиной: `wallHatchStepUnits(cellCm)` возвращает
`8 × (5 / cell_cm)` — это 9.6 см плана при любом масштабе сетки и ровно
исторические 8 юнитов при эталонном `cell_cm: 5`, так что старые планы не
двигаются. Толщина штриха следует за шагом, поэтому соотношение «штрих к
просвету» тоже перестало зависеть от масштаба.
Формула из описания issue (`cell_cm / 5`) не годилась: она увеличивает шаг там,
где стена и без того тонкая в юнитах, и разброс не исчезает, а растёт — 39
полос против 0.06 на краях диапазона. Множитель обратный; на эту ошибку
поставлен отдельный мутант `hatch-step-inverted`.
Зумовая компенсация `1/zoom` убрана по решению владельца (§4.2 ТЗ): стена,
которая меняет штриховку при зуме, — это тот же дефект, только по другой оси.
От каши на дальнем конце зума защищает второй порог `wallHatchNeedsSolid`
рядом с существующим `wallBodyNeedsSolid`; шаг клампится в [0.5, 80] юнитов,
чтобы патологический `cell_cm` не выродил паттерн.
Статический рендерер (`space-render.ts`) нёс собственную константу 8 и вообще
не знал про зум — то есть уже сегодня расходился с картой при любом зуме,
кроме единицы. Теперь оба читают шаг из одной функции; смок проверяет, что они
согласны между собой, а не только каждый сам с собой.
Два golden-эталона переснято осознанно (`golden:accept -- --reviewed`):
`large-house-zoom-040-dark` (шаг был 20 юнитов, стал 8) и
`large-house-zoom-250-dark` (был 3.2, стал 8). Расхождение осмотрено: меняется
только плотность штриховки тел стен, колонн и перегородок, геометрия и цвета
идентичны. Остальные 80 сцен не тронуты — `accept` переснимает весь набор, и
пять сцен, разошедшихся на шуме рендера, возвращены к прежним байтам вместе с
их хэшами в индексе.
Issue: #230
User-Visible: yes
Дефект High-1 был одинаковым в двух местах — и в живом рисовании, и в
«Оптимизировать планы», — потому что каждый вызывающий собирал геометрию
примыканий сам. Ревью r2 справедливо заметило, что и защита получилась
однобокой: юнит и мутант сторожили только оптимизатор, а путь карты — тот, где
дефект и был виден пользователю, — не сторожил никто. Заплатка в виде второго
мутанта-близнеца оставила бы причину на месте: два списка координат, которые
обязаны совпадать, но ничем не связаны.
Поэтому геометрия переехала в `spaceMergeGeometry(space, { excludeDraftId })`:
один источник комнат, колонн и концов черновиков, одни координаты, одно место,
где можно ошибиться. Оба вызывающих теперь строчка вызова.
Покрытие идёт за причиной, а не за симптомом: три юнита в
`test/wall-merge.test.mjs` проверяют масштаб полигонов (включая комнаты в форме
x/y/w/h и комнату без геометрии), исключение активного черновика и сам T-стык к
середине стороны комнаты. Мутанты `partition-merge-rescales-rooms` и
`chain-merge-sees-own-draft` перенацелены на общий модуль и теперь краснеют для
обоих путей сразу: 2 и 1 падение, проверено применением патча.
Сценарий с комнатой в смоке пробовал — не взлетел: рисование в комнату
поднимает `_offerWallFaces`, и цепочка не завершается штатно. Ломать смок под
тест не стал, юниты общего модуля покрывают оба пути честнее.
Issue: #229
User-Visible: no
**High-1.** Комнаты хранятся в тех же координатах, что и перегородки:
`roomPoly` отдаёт сырой полигон конфига. Обе обвязки делили его на `NORM_W`
ещё раз, комната уезжала в область ~0.0001, и `junctionAt` не находил ни
одного совпадения. Узел на T-стыке к середине стены комнаты — тот самый
случай, ради которого ТЗ прошло два раунда ревью, — молча исчезал.
Воспроизведено вызовом `optimizePlans`: `partitionsMerged === 1` там, где
ожидается 0.
**High-2.** Завершаемая цепочка к моменту слияния ещё лежит в `room_drafts`:
каждый клик персистит её через `_persistActiveDraftSegment`, а удаляется
черновик строкой ниже вызова слияния. Собственные концы цепочки считались
чужим примыканием, и стык с существующей стеной не срастался. Активный
черновик теперь исключается — ровно так же, как это делает
`plan-snap-overlay` (`activeDraftId`).
Дыры в тестах, которые это пропустили, закрыты по существу, а не заплаткой:
- `demo/smoke_wall_chain_merge.mjs` рисует продолжение реальными кликами через
`_markupClick`, а не присваиванием `_path`, — то есть исполняет тот путь, на
котором дефект и жил. Клики задаются в координатах плана и переводятся через
живой view box, иначе смок целится мимо только что нарисованной стены.
- `test/plan-optimizer.test.mjs` получил комнату с примыканием к середине
стороны: юниты модуля этого не ловили, потому что передают полигон уже в
согласованном масштабе, минуя обвязку.
- Мутанты `partition-merge-rescales-rooms` и `chain-merge-sees-own-draft`
сторожат оба места: проверены применением патча, 1 и 2 падения.
Issue: #229
User-Visible: no
`scripts/check-docs.mjs` держит скриншоты в соответствии с исходниками через
отпечаток `src/**`. Отпечаток был просрочен ещё до этой задачи — проверено
исполнением на чистом dev, — так что перезахват закрывает чужой долг заодно с
изменением, которое отпечаток всё равно бы сдвинуло. Содержательно кадры те
же: меняется только шум рендера.
Issue: #229
User-Visible: no
Рисование прямой стены в несколько кликов оставляло по записи на каждый
отрезок. Швы невидимы, пока их не тронешь: выделение хватает кусок,
перетаскивание ломает стену пополам, толщина задаётся пофрагментно. У стен
комнат этого давно нет — `normalizeWallIntervals` схлопывает каждый сплошной
участок одной толщины. Независимые перегородки жили по другому правилу.
Новый чистый модуль `src/wall-merge.ts` даёт им то же правило:
- `mergeCollinearPartitions` сращивает соседей одинаковой толщины и
направления до неподвижной точки, но только там, где узел никому не нужен.
Узел остаётся, если в него приходит третья перегородка, стена комнаты
(стороной, а не только вершиной), колонна или конец сохранённого черновика.
- Направление выжившей записи канонизируется лексикографически: иначе одна и
та же физическая стена выходила то a→b, то b→a в зависимости от порядка
входа, и каждый host.t вдоль неё переворачивался вместе с ней.
- `applyOpeningMoves` переносит проёмы на выжившую запись: и авторитетный
`host`, и legacy-проекцию `x/y/angle`, которую рисует старый читатель
конфига (docs/CONFIG-COMPATIBILITY.md, #132). Проекция здесь не кэш —
канонизация направления разворачивает угол на 180°.
Рисование сращивает только свою цепочку и то, чего она коснулась (§8.6 ТЗ):
молча править чужие швы в стороне оно не вправе — для этого есть
«Оптимизировать планы» с предпросмотром, отчётом и отменой. Оптимизация
проходит по всему пространству без seed-ограничения и отдельной строкой
сообщает, сколько записей исчезло.
Issue: #229
User-Visible: yes
The tolerance fix in r2 named the room's nearest polygon vertex as the point
of contact, which silently excluded the T-junction — a partition meeting the
middle of a room wall. That is a documented product case (141-wall-junctions
§13.1) and the code already measures distance to the edge, not the vertex
(distToSegment over roomEdges). Merging would have run straight through a
legitimate node.
AC2 now proves the room case with a T-junction into the middle of a long
side, and a mutant restores the vertex-only search.
Issue: #229
User-Visible: no
M1: "merge across the whole space when a chain ends" quietly overreached the
owner's split — drawing fixes its own seam, the optimiser fixes what has piled
up, and only the latter comes with a report and an undo. Section 8.6 now scopes
it to the connected component the new chain belongs to.
M2: the tolerance for "something else meets here" was only defined for a
partition-to-partition joint. Room edges, columns and drafts now use the same
EPS_JOIN — a gap cannot be a junction in one case and not in another.
M3: an opening also carries a materialised x/y/angle projection that
CONFIG-COMPATIBILITY (#132) requires to stay in step with its host, and the
code already re-materialises it after every host change. Merging is such a
change; AC3 now fails if only host.t is recomputed and the projection goes
stale.
Issue: #229
User-Visible: no
Written on the owner's decisions of 2026-08-21: merge as the chain is
finished, sweep already-drawn plans from "Optimise plans", and keep merging
even when an opening sits on the seam.
The part that is easy to miss is that last one. An opening stores its
position as a fraction of its host's length, so merging two partitions
changes the length under it and moves the door unless the fraction is
recomputed. AC3 therefore checks the door's coordinates in plan units, not
that a field was rewritten.
Issue: #229
User-Visible: no
Three code-review rounds on #220 published a verdict and then failed the
run: the document never reached the branch, so the #171 guard refused
before the label step and neither the merge nor S8-merged happened. The
cause was structural. The document lived as an untracked file inside the
very checkout the reviewer edits while proving that a test can fail, and
restoring that tree — git checkout, git clean — deletes an untracked file.
Spec rounds survived only because they never mutate anything.
The reviewer now writes to REVIEW_DOC under RUNNER_TEMP, outside the
repository, and the publish step copies it into docs/reviews before
committing. Tree cleanup can no longer destroy the artefact, and the
reviewer no longer needs to touch docs/reviews at all.
Verified against a local git fixture on five paths: document outside the
repo with a mutated tree, nothing anywhere (loud failure), document only in
the working copy, document already committed by the reviewer, and a branch
that moved during the review.
Same file as main, byte for byte.
Issue: #220
User-Visible: no
Review CODE-REVIEW-220-r2/r3, F1.
The M1 fix installs window listeners for the length of the gesture, and
disconnectedCallback — which takes down everything else, down to the other
local gesture — did not take those down. Losing the card mid-drag (Lovelace
rebuilding its tree, the user leaving the view with the button still down)
left them alive: the closure holds the instance and its config, and the next
pointerup anywhere on the page would have an invisible card write its order.
The smoke now holds a tab, removes the card, and checks that the release it
should no longer hear changes nothing. Registered as a mutant too.
Issue: #220
User-Visible: no
Review CODE-REVIEW-220-r1.
H1: markersNeedingPlacement decided who depends on the order by reading
marker.area alone, while resolveExplicitMarkerPlacement reads
`marker.area || <area of the HA device>`. The ordinary marker — bind an HA
device, store neither field — is anchored by the registry and never depended
on the order, yet it was being written a space it never asked for. Dormant
today, and the day that HA area changes it moves the marker to whatever space
used to be first. The resolver now asks for the area actually in force.
M1: a mouse released past the panel left the gesture stuck, swallowing the
next click. Pointer capture is the usual answer and is now taken, but it is
not a guarantee — the browser grants it only for a live pointer. The window
listener is what actually closes the gesture.
M2: the fifth mutant from the spec is registered, plus a sixth for the stuck
drag above.
Writing the smoke for M1 turned up why the first attempt passed against
broken code: synthetic PointerEvents default to composed:false and never
leave the shadow root, so nothing outside the panel could ever hear them.
Real pointer events are composed; the smoke now says so.
Issue: #220
User-Visible: no
The order of config.spaces used to be whatever order the spaces were created
in, and there was no way back other than deleting a space and drawing it
again.
The gesture is deliberately narrow — mouse, editors only. The same tabs are
the primary way to switch spaces in View, where touch is first class, so a
drag there would compete with the tap that switches. Recorded in the spec as
"Touch editor: not exposed".
The part that needed care is not the drag. Position in the array feeds three
things: the marker placement fallback, the swipe neighbour and a positional
`floor`. So the write that stores the new order also writes down the
placement that used to depend on it: a marker with neither an explicit space
nor an area that names one gets the space it has right now. Both changes go in
one save; splitting them would leave a window in which markers move on their
own. The positional `floor` cannot be fixed from here, so the card says so
once.
Issue: #220
User-Visible: yes
Section 4 now says what the pipeline does: a cycle is a verdict with
blocking findings followed by a return to the author, so a green verdict
consumes nothing — the case that cost #225 an arbitration after a failed
merge forced a rebase and a third attempt.
The canon also separates the two quantities the verdict line carries. The
attempt number names the review document, because two runs sharing a
number would overwrite each other's artefact; the budget counts blocking
cycles only. That is why the document threshold in the process gate sits
above the cycle limit, and why the guard reports a recount of review-4
rather than removing the label itself.
Issue: #227
User-Visible: no
The pipeline punished what it prescribed: after a failed merge it tells the
author to rebase and restore S7-code-review, and that attempt finished the
budget. On #225 a green code review with green CI ended in review-4.
Only yellow and red verdicts spend the budget now; a green verdict returned
nothing and consumes nothing. Attempts and cycles became separate
quantities: the attempt number names the review document, the limit
compares blocking cycles. The exhaustion comment lists what it counted, and
the guard reports a recount instead of stripping review-4 on its own.
Same file as main (41325a8), byte for byte.
Issue: #227
User-Visible: no
The pipeline punished what it prescribed: after a failed merge it tells the
author to rebase and restore S7-code-review, and that attempt finished the
budget. On #225 (light track, limit 2) the sequence yellow, green, rebase
produced review-4 on a task whose code review was green and whose CI was
green, with no product change after the verdict — the owner had to
arbitrate work that was already accepted.
A cycle under section 4 is a verdict with blocking findings followed by a
return to the author, so only yellow and red verdicts spend the budget now.
A green verdict returned nothing and consumes nothing, which also removes
any need to mark rebase re-runs specially.
Attempts and cycles are now separate quantities. The attempt number keeps
naming the document, because two runs sharing a number would overwrite each
other's review artefact, while the limit compares blocking cycles only. The
exhaustion comment lists the verdicts it counted, and the guard no longer
strips review-4 — it reports the recount and leaves the decision with the
owner.
Rule 7 of the process gate follows: its document threshold rises above the
cycle limit, because legitimate attempts can exceed cycles and a threshold
equal to the limit would refuse the very rebase the pipeline demands.
Issue: #227
User-Visible: no
The assumptions section is explicitly labelled "technical, free to change",
and it held a requirement that AC3 and a mutant already test as a fact. Read
literally, it invited splitting the write in two — reopening the very window
in which markers move. The point now states the opposite: everything else in
that section is free, this one is normative and lives in section 8.3.
Issue: #220
User-Visible: no
M1: the spec now carries the touch classification TOUCH-SUPPORT.md asks every
editor feature for — "Touch editor: not exposed", with the reason it is a
decision rather than an omission.
M2: the first draft denied adding a config field in one section while planning
to store an anchor in settings in another. Resolved by dropping the anchor:
reordering materialises the placement that was implicit, giving those markers
an explicit space in the same write. No new field, no schema change, and the
marker stays exactly where the user saw it.
Issue: #220
User-Visible: no
Written on the owner's product decisions of 2026-08-20: mouse only and only in
the editor modes, one warning about the positional `floor` from #210, no
keyboard alternative.
The spec carries the part that is easy to miss — the order of `config.spaces`
is not decoration. It feeds the marker placement fallback, the swipe
neighbour and the numeric `floor`, so reordering tabs must not move a single
marker. That is a named acceptance criterion with a mutant behind it.
Issue: #220
User-Visible: no
Import of a backup holding PDF attachments: the content resolver parses a url
as a url, and the three mutants guarding it are registered. The user-visible
change is documented in 4a84734, which carries both changelog entries — this
merge adds no behaviour of its own.
Code review r2 green (docs/reviews/CODE-REVIEW-225-r2.md). The third pass was
a rebase over #226, not a fix — owner arbitration on the review-4 the cycle
counter raised for it (PROCESS.md §4; counter defect filed as #227).
Issue: #225
User-Visible: no
Review CODE-REVIEW-225-r1.
M1: urlsplit(url).path was trusted even when the url carried a scheme or an
authority, so "https://evil.example/houseplan_files/files/m1/doc.pdf"
resolved onto a local file while _looks_internal kept calling it external —
the mirror image of the inconsistency this resolver exists to prevent. Only a
same-document reference is resolved by its path now.
M2: the three mutants the spec described are registered in
scripts/mutation-gate.mjs instead of living as a one-off manual run. The
traversal entry drops both structural checks at once on purpose: taken one at
a time the defence is layered (sanitize_marker_id turns ".." into "misc") and
the mutant would be equivalent — established by running it.
Issue: #225
User-Visible: no
A backup holding a PDF attachment could not be imported back: legacy links
carry a cache-buster (".../files/m1/doc.pdf?v=1783170649"), and the resolver
compared the raw tail with its sanitized form, so the query made the name
differ from itself. The reference then read as internal by prefix and
non-canonical by name, which is exactly the combination _content_state must
refuse — every such document failed with invalid_content.
Parse the url as a url: the path addresses the file, the query and the
fragment address the transfer. Path segments keep doing the guarding, so
dropping the query cannot widen what a segment is allowed to be.
Issue: #225
User-Visible: yes
The reviewer prompt was identical for every round, and the canon said
nothing about the scope of a repeat pass, so r2 re-derived the product
framing and re-checked acceptance criteria the fix never touched: the r2
pass on #150 cost a full pipeline run over one line in a test fixture.
From the second cycle on, the subject is the delta against the SHA the
previous verdict was given on: each earlier finding must be shown closed
by a line of code or text, only the criteria the delta can reach are
re-verified, and whatever is carried over is listed with the round and SHA
it came from. Cheap gates still run every round.
The scope shrinks, the strictness does not. A fix can break a criterion an
earlier round accepted — that is how regression #102 happened — so the
boundary is the findings plus everything the delta can reach, and a
non-local delta (a rebase onto a moved dev, a behaviour contract change, a
new subsystem) still gets the full pass.
Issue: #214
User-Visible: no
The reviewer prompt was identical for every round, and the canon said
nothing about the scope of a repeat pass, so r2 re-derived the product
framing and re-checked acceptance criteria the fix never touched: the r2
pass on #150 cost a full pipeline run over one line in a test fixture.
From the second cycle on, the subject is the delta against the SHA the
previous verdict was given on: each earlier finding must be shown closed
by a line of code or text, only the criteria the delta can reach are
re-verified, and whatever is carried over is listed with the round and SHA
it came from. Cheap gates still run every round.
The scope shrinks, the strictness does not. A fix can break a criterion an
earlier round accepted — that is how regression #102 happened — so the
boundary is the findings plus everything the delta can reach, and a
non-local delta (a rebase onto a moved dev, a behaviour contract change, a
new subsystem) still gets the full pass.
Issue: #214
User-Visible: no
Every push to dev paid for the full browser trio and the backend suite,
including commits that touch only documentation, workflows or process
scripts — the bundle and the harness were byte-identical, so the runs
proved nothing new. On 2026-08-19 alone that was roughly six pushes at
about seven minutes each.
The reuse key per heavy job is sourceFingerprint (src, demo fixtures,
golden scenarios, build manifests) plus that job's own harness: smoke
takes demo/smoke_*.mjs, golden takes demo/golden/** including baselines,
performance_smoke takes demo/performance/**, backend takes tests_backend
and the Python sources. A cache marker is written only by a successful run
of the same key, so a hit proves a job with identical inputs already
passed. scripts/** is deliberately outside every key: infrastructure work
edits it constantly and reuse would never fire.
This is not the path filter from the `changes` job, which stays disabled
on dev on purpose: there the scope is guessed from paths and "green" means
different things, here input equivalence is proven by a hash. And a
release candidate always bumps the version, which is part of the
fingerprint, so its keys are new by construction and the full gate set
still runs before every beta and release.
A waived job is announced with a notice and a run summary line rather than
skipped in silence, and the marker save tolerates a concurrent identical
run instead of reddening the job.
Issue: #208
User-Visible: no
Rule 8 demanded a working S-label from every class A/B commit's issue,
while owner decision #118 sends infrastructure work outside the S1..S8
flow entirely — such an issue has no status label by construction. The
two rules contradicted each other and the machine-checked one won, so
Validate on dev went red on every infrastructure commit (#175, #191,
#202, #206) and the catch-up signal stopped meaning anything. A gate that
is always red is not a gate.
The waiver keys on the diff, not on a permission label: a range with no
class A file at all. An `infra` label could be pinned on a product task
to walk a product commit past the status check; ceasing to touch class A
without ceasing to be infrastructure work is not possible. Issue
existence, open state, `blocked` and fail-closed on an unreachable gh all
still apply, and the waiver prints a visible warning rather than passing
in silence.
Mutation-checked both ways: unwiring the waiver reddens the CLI test,
and letting class A keep the waiver reddens both new tests.
Issue: #207
User-Visible: no
performance_smoke burned nearly all of its 15-minute budget before the
benchmark even started, twice in a row: validate.yml had no browser cache
at all, so every browser job paid for a full `playwright install
--with-deps` — apt work the ubuntu-latest image makes redundant, with
unbounded retries against an unreachable azure mirror on top. For a
measuring job that is worse than lost minutes: the timing window competes
with package installation on the same runner.
#175 fixed this for the review pipeline but deliberately left the flag
here, reasoning that a prerelease gate values predictability over
minutes. That reasoning was wrong — the flag is what made the gate
unpredictable.
Browsers are now cached per package-lock hash in smoke, golden,
performance_smoke and the full performance run; installation happens only
on a cache miss and no longer touches apt. performance_smoke keeps
headroom for a cold cache at 20 minutes. If the image ever drops a
required library, Chromium fails to launch with a clear missing-libraries
error; that is the moment to bring the flag back.
Issue: #206
User-Visible: no
Filing and servicing a separate issue costs far more than fixing a small
problem in place — the owner's call of 2026-08-19 (#202). A Medium finding
inside the task's scope no longer becomes its own issue: with no High
findings the verdict is yellow, the author fixes it and the fix passes
another review cycle. Only an out-of-scope Medium is still filed
separately, because foreign scope is never patched from a task branch.
Applied to the canon (PROCESS.md), the reviewer prompt in process.yml and
AGENTS.md; the verdict format now writes "Medium: N -> in-task | #NN".
Issue: #202
User-Visible: no
Filing and servicing a separate issue costs far more than fixing a small
problem in place — the owner's call of 2026-08-19 (#202). A Medium finding
inside the task's scope no longer becomes its own issue: with no High
findings the verdict is yellow, the author fixes it and the fix passes
another review cycle. Only an out-of-scope Medium is still filed
separately, because foreign scope is never patched from a task branch.
Applied to the canon (PROCESS.md), the reviewer prompt in process.yml and
AGENTS.md; the verdict format now writes "Medium: N -> in-task | #NN".
Issue: #202
User-Visible: no
The old fixture ran the branch into the end of p1, so the T-patch lived
beyond the host (x>100) and the probe point [92,0] sat outside it under
any code behaviour: deleting the cutPartitionBody flatMap over patches
kept all subtests green (#188, found at the #132 code review).
The branch now meets the middle of the span, putting both node patches
inside the default opening's slot. The test proves its own fixture first:
an uncut run must show the patches bridging the slot, so if the geometry
ever stops producing them the control goes red instead of silently
devaluing the real assertion. Mutation-checked: reverting the flatMap
fails exactly this test, 888/889.
Issue: #188
User-Visible: no
After the mandatory rebase of a published issue branch the pre-push hook
still passes remote_old..local_new, and once the old tip is no longer an
ancestor that range drags in the whole advanced dev history: on #117 it
meant 84 foreign commits and 20 false rule-8 rejections over already
closed issues, leaving --no-verify as the only exit.
The clamp lives in the gate rather than the hook: .githooks/pre-push
carries an executable bit that MCP publication strips (the commit-msg
precedent), so editing it needs an owner-side commit. When the target is
an issue branch and the declared base is not an ancestor of the head, the
base becomes the merge-base with origin/dev. Fast-forward pushes keep
their exact range, every own commit is still judged, and a real violation
in a post-rebase commit still blocks — covered by a scenario test that
goes red without the wiring.
Issue: #190
User-Visible: no
The canonical job list predated the `docs` job (added 2026-08-16) and
omitted `process-gate`. `docs` is a real blocking gate — its fingerprint
check went red right after the #113 merge and cost an extra review cycle
of confusion. The list now matches validate.yml and names `changes` as a
service path-filter rather than a gate.
Issue: #191
User-Visible: no
On a Playwright cache miss the flag pulled Chromium's system libraries
through apt, spending minutes of the 45-minute review budget on packages
the ubuntu-latest image already ships — and the runner's retries against
the unreachable azure mirror made the step look hung on a live run. If the
image ever drops a required library, Chromium fails to launch with a clear
missing-libraries error; that is the moment to bring the flag back.
validate.yml keeps the flag deliberately: it is the prerelease gate, where
predictability is worth more than minutes.
Issue: #175
User-Visible: no
On a Playwright cache miss the flag pulled Chromium's system libraries
through apt, spending minutes of the 45-minute review budget on packages
the ubuntu-latest image already ships — and the runner's retries against
the unreachable azure mirror made the step look hung on a live run. If the
image ever drops a required library, Chromium fails to launch with a clear
missing-libraries error; that is the moment to bring the flag back.
validate.yml keeps the flag deliberately: it is the prerelease gate, where
predictability is worth more than minutes.
Issue: #175
User-Visible: no
On #150 both spec-review verdicts survived only as issue comments: the
publish step found nothing staged, printed a warning, and exited zero, so
the label moved and the missing artifact went unnoticed until the next
review caught it (#171). A verdict without a document in docs/reviews/ now
fails the run before the label step, preserving the invariant that an
unchanged label means a failed run.
An empty working copy alone is not a failure: the reviewer occasionally
commits the document itself through its app token, bypassing this step
(CODE-REVIEW-150-r1, committer GitHub), so the branch is checked first. A
postcondition verifies the exact expected filename reached the branch, and
the rebase-conflict path no longer exits zero either.
Issue: #171
User-Visible: no
On #150 both spec-review verdicts survived only as issue comments: the
publish step found nothing staged, printed a warning, and exited zero, so
the label moved and the missing artifact went unnoticed until the next
review caught it (#171). A verdict without a document in docs/reviews/ now
fails the run before the label step, preserving the invariant that an
unchanged label means a failed run.
An empty working copy alone is not a failure: the reviewer occasionally
commits the document itself through its app token, bypassing this step
(CODE-REVIEW-150-r1, committer GitHub), so the branch is checked first. A
postcondition verifies the exact expected filename reached the branch, and
the rebase-conflict path no longer exits zero either.
Issue: #171
User-Visible: no
Issue #150 reached a green verdict and then hit two pipeline defects at once.
The review document push came back 403 as github-actions[bot]: the PAT had
died, and checkout's persisted credential quietly took its place — a masked
actor instead of a loud failure. Credentials are no longer persisted, and the
token is now proven alive before the review starts, not after forty minutes of
reviewer work.
Branch selection took the first match alphabetically, and with a spec-era
branch sitting next to the implementation branch that meant the stale one.
The freshest branch by commit date is chosen instead, with a warning naming
every candidate when more than one exists.
Verified against the real #150 branches: the fix branch wins, the warning
fires.
Issue: #114
User-Visible: no
Issue #150 reached a green verdict and then hit two pipeline defects at once.
The review document push came back 403 as github-actions[bot]: the PAT had
died, and checkout's persisted credential quietly took its place — a masked
actor instead of a loud failure. Credentials are no longer persisted, and the
token is now proven alive before the review starts, not after forty minutes of
reviewer work.
Branch selection took the first match alphabetically, and with a spec-era
branch sitting next to the implementation branch that meant the stale one.
The freshest branch by commit date is chosen instead, with a warning naming
every candidate when more than one exists.
Verified against the real #150 branches: the fix branch wins, the warning
fires.
Issue: #114
User-Visible: no
The previous merge (6beb404) took a stale local ref and brought only the
specification; this one brings the implementation and the error-channel fix.
Issue: #164
User-Visible: no
The analyst used to ask the owner to confirm every estimate and waited for an
answer on each point. Most issues are unambiguous, and most of that waiting
changed nothing — the owner's own measure is that seven issues in ten should
travel from S1-new to a finished spec without a single question.
Section 2.2 flips the default. Estimates, type, priority and track go on as
labels immediately; the analysis comment is a notification, not a request —
the owner's silence is consent, his disagreement is a label edit, and neither
stops the work. The analyst moves the issue to S3-spec himself. The only
questions that ever reach the owner are product questions, asked at the spec
stage in one batch with defaults and blocked, and only when the spec cannot be
written without the answer; anything that can wait for the spec waits, anything
that does not block it becomes a recorded assumption instead. The one full stop
left in analysis is a genuine SCOPE conflict, where the analyst proposes
rejection and the owner decides.
Issue: #114
User-Visible: no
The author agent read STATUS.md, saw the owner's 2026-08-07 rule that ordinary
fixes are made locally without tests or commits, correctly ranked it below
AGENTS.md and PROCESS.md, and followed the canon instead. That is the trust
order doing its job — and the canon's second half says a divergence is not
ignored but fixed.
The line now says what replaced it: since release 1.62 every product change goes
through the process — an issue in S5-ready or later, a task branch, trailers on
every commit, the review pipeline. The release mechanics in the same cell were
still accurate and stay.
Issue: #114
User-Visible: no
The owner's machine now carries Playwright with Chromium on Windows and a full
WSL environment — verified by execution: 34/34 smoke assertions, and 242 backend
tests passed where native Windows silently skips every test_ha_* file. A red
smoke that reaches the review costs a cycle of forty-five minutes plus the
return trip; run locally it costs a minute, and #89 already paid that price
once.
WSL runs of the full harness and golden verify are advisory. The canon does not
move: the beta gate is CI at the exact SHA, and baselines are accepted only via
golden:accept --reviewed on a complete Linux CI artefact.
Issue: #151
User-Visible: no
First real run of the gate failed before reaching a single mutant: the clean
run of the golden guard was red on untouched code. demo/golden/policy.mjs
refuses `verify --scenario=...` on purpose — a partial verify is the "make CI
green" loophole the policy exists to close. The gate built to catch dishonest
tests had reached for a dishonest shortcut, and the policy caught it.
capture keeps the whole check: a failed semantic assertion becomes status
error, and goldenRunFailed treats an error as failure in either mode. The scene
carries warmPixelRegion with minPixels 2500 over the receiving half, so a lamp
moved out of reach still fails it — which is exactly what this mutant asserts.
Issue: #85
User-Visible: no
gh workflow run answered 404: workflow_dispatch and schedule both resolve the
workflow file against the default branch, and the file sat only in dev. The
same trap as the process pipeline — even documented in that file's header — and
still stepped in a second time. The job itself checks out dev, so running from
main tests exactly the code it should.
Issue: #85
User-Visible: no
Two agents sharing one checkout share one HEAD, and twice in an hour a commit
landed on someone else's task branch that way. The layout that ends it: the main
clone belongs to the author and its task branches, hp-dev is the owner's
permanent worktree on dev, and the reviewer and the infrastructure agent own no
local tree at all — one runs in CI on a fresh checkout, the other reads through
git show and publishes through the API, so it has no HEAD to collide with.
Also recorded: a worktree is only usable on the machine that created it, because
its .git file stores an absolute path in that machine's format. We hit this in
both directions within a day.
Issue: #115
User-Visible: no
On Windows URL.pathname yields /C:/..., which spawnSync then reads as C:\C:\...
and the whole npm test run dies in this one test. Linux CI never caught it
because both spellings coincide there — which is exactly why the canonical gate
lives on Linux and the local run is advisory.
Issue: #133
User-Visible: no
One word was mistyped while transferring the file: "на каждый HA state
update" instead of "на каждом". The moved document must match the original
byte for byte.
Issue: #142
User-Visible: no
Three review documents sat in the repository root, committed before the pipeline
existed and before docs/reviews/ did. The directory exists now and the pipeline
writes into it, so they move there and the root stops being a second place to
look.
The #89 spec had a twin: the research draft next to the normative stage1
document, two files for one issue. The draft goes to legacy — it fed the
decisions and is worth keeping, but nothing should read it as current.
ROADMAP.md carried a live link to the Project v2 board that was dropped
yesterday; missed then because the sweep grepped for status-canon wording, not
for every link. And docs/README.ru.md said "verified against v1.60.0" as if
that were fresh — the line is now an explicit warning naming what to trust
instead: USER-GUIDE.ru.md and the changelogs.
Issue: #142
User-Visible: no
Three lines of section 10.4 and the trailing newline went missing in transit.
The lost paragraph is the one that says a label which did not change means the
run failed rather than the work — the sentence that tells a waiting author to
read the logs instead of polling for another forty-five minutes. Losing exactly
that one while copying a document about silent failures is a joke the situation
made on its own.
Caught by the byte comparison that follows every publish, which is the whole
reason it follows every publish.
Issue: #139
User-Visible: no
The owner stopped using GitHub Projects. Most of this is wording, but one part
was not: release-prerelease.mjs talked to the Project in code. finishIssues
looked up the project id, listed its items and its Status=Done option, and threw
when an issue was missing from the board — so the first release that closed an
issue would have died on a step with nothing to do with publishing. Found by
reading rather than by releasing, which was luck.
Closing issues stays, and now strips the status label first. That order is not
cosmetic: the invariant that a closed issue carries no status label has broken
twice already, both times because a manual step did it the other way round. The
close-merged job already does it in this order.
The documents now say labels and only labels. The explicit "no longer used"
lines are kept on purpose, in PROCESS.md and next to the code that used to sync:
a decision that vanishes quietly gets reintroduced a month later by someone who
never knew it was made.
Issue: #139
User-Visible: no
The owner stopped using GitHub Projects. Most of this is wording, but one part
was not: release-prerelease.mjs talked to the Project in code. finishIssues
looked up the project id, listed its items and its Status=Done option, and threw
when an issue was missing from the board — so the first release that closed an
issue would have died on a step with nothing to do with publishing. Found by
reading rather than by releasing, which was luck.
Closing issues stays, and now strips the status label first. That order is not
cosmetic: the invariant that a closed issue carries no status label has broken
twice already, both times because a manual step did it the other way round. The
close-merged job already does it in this order.
The documents now say labels and only labels. The explicit "no longer used"
lines are kept on purpose, in PROCESS.md and next to the code that used to sync:
a decision that vanishes quietly gets reintroduced a month later by someone who
never knew it was made.
Issue: #139
User-Visible: no
The owner stopped using GitHub Projects. Most of this is wording, but one part
was not: release-prerelease.mjs talked to the Project in code. finishIssues
looked up the project id, listed its items and its Status=Done option, and threw
when an issue was missing from the board — so the first release that closed an
issue would have died on a step with nothing to do with publishing. Found by
reading rather than by releasing, which was luck.
Closing issues stays, and now strips the status label first. That order is not
cosmetic: the invariant that a closed issue carries no status label has broken
twice already, both times because a manual step did it the other way round. The
close-merged job already does it in this order.
The documents now say labels and only labels. The explicit "no longer used"
lines are kept on purpose, in PROCESS.md and next to the code that used to sync:
a decision that vanishes quietly gets reintroduced a month later by someone who
never knew it was made.
Issue: #139
User-Visible: no
The owner stopped using GitHub Projects. Most of this is wording, but one part
was not: release-prerelease.mjs talked to the Project in code. finishIssues
looked up the project id, listed its items and its Status=Done option, and threw
when an issue was missing from the board — so the first release that closed an
issue would have died on a step with nothing to do with publishing. Found by
reading rather than by releasing, which was luck.
Closing issues stays, and now strips the status label first. That order is not
cosmetic: the invariant that a closed issue carries no status label has broken
twice already, both times because a manual step did it the other way round. The
close-merged job already does it in this order.
The documents now say labels and only labels. The explicit "no longer used"
lines are kept on purpose, in PROCESS.md and next to the code that used to sync:
a decision that vanishes quietly gets reintroduced a month later by someone who
never knew it was made.
Issue: #139
User-Visible: no
Every push to every branch ran 128 browser smokes, 50 golden scenes, a Home
Assistant install and a performance pass — including a push that added one spec
file. The pipeline made such pushes routine: every spec revision and every
review document is a push to a task branch and used to cost the full suite.
A changes job classifies the push range; frontend, smoke, golden, performance
and backend now run only when their paths moved, and hacs and hassfest only for
manifests, translations or Python. provenance and process-gate always run — they
judge commits, not code.
The exception carries the design. On dev everything runs, always, unfiltered:
the beta gate accepts "green Validate at the exact SHA", and if the volume of a
run depends on the diff, green stops meaning one thing — a release candidate
touches manifests and changelogs, would skip the browser suites under filtering,
and a run with skipped jobs still concludes success. That would be the sixth
silent success of the week. Filters save time on task branches, where Validate is
an early signal and the real acceptance is the code review running gates itself.
A new branch with a zero before-sha is classified from the merge-base with dev,
not from the root of history.
Issue: #136
User-Visible: no
The file declared itself pure but imported the module through the package, and
the package __init__ unconditionally imports homeassistant. Without homeassistant
installed pytest did not skip the file — it stopped collecting the whole
tests_backend directory, taking the previously working pure suite down with it.
test_validation.py had already established the by-path pattern; virtual_lights.py
imports nothing beyond the standard library, so it loads cleanly.
The async tests also dropped their pytest-asyncio dependency in favour of
asyncio.run: the offline environment does not carry the plugin, and without it
the two tests failed as unsupported async defs. The offline gate has to be green,
or nobody runs it.
Verified in both environments: pytest+voluptuous only — 129 passed where
collection previously stopped dead; with pytest-asyncio as in CI — 129 passed.
Issue: #135
User-Visible: no
Five times in this project a green test meant nothing was checked. The
continuity smoke stayed green after the entire mechanism it guards was cut out.
The golden scene created to protect doorway light was empty — 1,177 warm pixels
against 107,119, all of them icons. The shadow smoke passed while no shadow was
drawn. Each time the test had been written alongside the code, went green at
once, and nobody ever asked whether it could go red.
The gate makes that question routine. Each mutant is a few lines of patch that
reproduce a known breakage, plus the name of the test that must fail on it. A
worktree is patched, the bundle rebuilt, the guard run — and a guard that stays
green fails the gate. Six mutants cover the holes documented in #85; the anchors
are exact strings from today's source, so the registry cannot silently drift —
a unit test that runs with the ordinary suite refuses a stale anchor.
The full run rebuilds the bundle per mutant, so it lives in its own workflow,
before a stable release and on a weekly schedule, not in Validate. The rules for
new tests are written at the top of docs/TESTING.md, and the sixth of them is
the cheapest: an assertion that reads back the property the code just set is
not written at all.
Issue: #85
User-Visible: no
Five times in this project a green test meant nothing was checked. The
continuity smoke stayed green after the entire mechanism it guards was cut out.
The golden scene created to protect doorway light was empty — 1,177 warm pixels
against 107,119, all of them icons. The shadow smoke passed while no shadow was
drawn. Each time the test had been written alongside the code, went green at
once, and nobody ever asked whether it could go red.
The gate makes that question routine. Each mutant is a few lines of patch that
reproduce a known breakage, plus the name of the test that must fail on it. A
worktree is patched, the bundle rebuilt, the guard run — and a guard that stays
green fails the gate. Six mutants cover the holes documented in #85; the anchors
are exact strings from today's source, so the registry cannot silently drift —
a unit test that runs with the ordinary suite refuses a stale anchor.
The full run rebuilds the bundle per mutant, so it lives in its own workflow,
before a stable release and on a weekly schedule, not in Validate. The rules for
new tests are written at the top of docs/TESTING.md, and the sixth of them is
the cheapest: an assertion that reads back the property the code just set is
not written at all.
Issue: #85
User-Visible: no
The owner's report: the process works but every stage takes a long time even on
simple bugs. Two causes, and neither was the one that first comes to mind.
The reviewer ran everything regardless. On #89 it installed Chromium, ran all 127
smoke files and a full golden capture — right for a task rated 10/10 for
complexity, absurd for a bug about a room divider. Full suites are the pre-beta
gate; the review now runs typecheck, unit and build always, and smokes, golden,
pytest or performance only where the diff and the AC call for them. The price of
narrowing it is honesty: the reviewer must list which gates it ran, which it did
not, and why, so a skipped gate is a visible decision rather than a silent one.
The reviewer also built its own environment out of model turns, with no npm cache
and no browser cache, paid for from the same forty-five minutes. The workflow now
installs dependencies and Chromium as ordinary cached steps, after switching to
the task branch so the lockfile is the branch's own.
Second, ceremony did not scale down. The light track makes a spec cheap; the new
trivial track does without one — S2-analysis straight to S5-ready, no spec review,
AC in the issue body. It is deliberately hard to qualify for: a bug on one surface,
no new UX contract, no migration, no i18n, no perf or touch effect, three checkable
AC at most, and expected behaviour already on record. Nothing left to decide is the
criterion that holds the whole thing up, and it cannot be met by feeling sure.
Code review is never skipped on either track. It is what stands in for testing
here, so it is the one stage speed may not buy.
Issue: #127
Issue: #128
User-Visible: no
The owner's report: the process works but every stage takes a long time even on
simple bugs. Two causes, and neither was the one that first comes to mind.
The reviewer ran everything regardless. On #89 it installed Chromium, ran all 127
smoke files and a full golden capture — right for a task rated 10/10 for
complexity, absurd for a bug about a room divider. Full suites are the pre-beta
gate; the review now runs typecheck, unit and build always, and smokes, golden,
pytest or performance only where the diff and the AC call for them. The price of
narrowing it is honesty: the reviewer must list which gates it ran, which it did
not, and why, so a skipped gate is a visible decision rather than a silent one.
The reviewer also built its own environment out of model turns, with no npm cache
and no browser cache, paid for from the same forty-five minutes. The workflow now
installs dependencies and Chromium as ordinary cached steps, after switching to
the task branch so the lockfile is the branch's own.
Second, ceremony did not scale down. The light track makes a spec cheap; the new
trivial track does without one — S2-analysis straight to S5-ready, no spec review,
AC in the issue body. It is deliberately hard to qualify for: a bug on one surface,
no new UX contract, no migration, no i18n, no perf or touch effect, three checkable
AC at most, and expected behaviour already on record. Nothing left to decide is the
criterion that holds the whole thing up, and it cannot be met by feeling sure.
Code review is never skipped on either track. It is what stands in for testing
here, so it is the one stage speed may not buy.
Issue: #127
Issue: #128
User-Visible: no
The owner's report: the process works but every stage takes a long time even on
simple bugs. Two causes, and neither was the one that first comes to mind.
The reviewer ran everything regardless. On #89 it installed Chromium, ran all 127
smoke files and a full golden capture — right for a task rated 10/10 for
complexity, absurd for a bug about a room divider. Full suites are the pre-beta
gate; the review now runs typecheck, unit and build always, and smokes, golden,
pytest or performance only where the diff and the AC call for them. The price of
narrowing it is honesty: the reviewer must list which gates it ran, which it did
not, and why, so a skipped gate is a visible decision rather than a silent one.
The reviewer also built its own environment out of model turns, with no npm cache
and no browser cache, paid for from the same forty-five minutes. The workflow now
installs dependencies and Chromium as ordinary cached steps, after switching to
the task branch so the lockfile is the branch's own.
Second, ceremony did not scale down. The light track makes a spec cheap; the new
trivial track does without one — S2-analysis straight to S5-ready, no spec review,
AC in the issue body. It is deliberately hard to qualify for: a bug on one surface,
no new UX contract, no migration, no i18n, no perf or touch effect, three checkable
AC at most, and expected behaviour already on record. Nothing left to decide is the
criterion that holds the whole thing up, and it cannot be met by feeling sure.
Code review is never skipped on either track. It is what stands in for testing
here, so it is the one stage speed may not buy.
Issue: #127
Issue: #128
User-Visible: no
The owner's report: the process works but every stage takes a long time even on
simple bugs. Two causes, and neither was the one that first comes to mind.
The reviewer ran everything regardless. On #89 it installed Chromium, ran all 127
smoke files and a full golden capture — right for a task rated 10/10 for
complexity, absurd for a bug about a room divider. Full suites are the pre-beta
gate; the review now runs typecheck, unit and build always, and smokes, golden,
pytest or performance only where the diff and the AC call for them. The price of
narrowing it is honesty: the reviewer must list which gates it ran, which it did
not, and why, so a skipped gate is a visible decision rather than a silent one.
The reviewer also built its own environment out of model turns, with no npm cache
and no browser cache, paid for from the same forty-five minutes. The workflow now
installs dependencies and Chromium as ordinary cached steps, after switching to
the task branch so the lockfile is the branch's own.
Second, ceremony did not scale down. The light track makes a spec cheap; the new
trivial track does without one — S2-analysis straight to S5-ready, no spec review,
AC in the issue body. It is deliberately hard to qualify for: a bug on one surface,
no new UX contract, no migration, no i18n, no perf or touch effect, three checkable
AC at most, and expected behaviour already on record. Nothing left to decide is the
criterion that holds the whole thing up, and it cannot be met by feeling sure.
Code review is never skipped on either track. It is what stands in for testing
here, so it is the one stage speed may not buy.
Issue: #127
Issue: #128
User-Visible: no
Issues labelled before the pipeline existed keep their spec straight in dev and
have no issue/NN branch. The publish step quietly exited zero for them, so the
verdict would arrive as a comment and the analysis behind it would be thrown
away — the fifth instance today of a step reporting success by doing nothing.
The document now goes wherever the spec itself lives: the task branch when there
is one, dev otherwise. Publishing also survives dev moving on while the review
ran, which takes up to forty-five minutes, by rebasing once before it gives up.
Four issues are waiting on this — #12, #30, #44 and #52 — each with a spec in dev,
a status label applied during the bulk pass in August and a review that never ran
because nothing was there to raise the event.
Issue: #114
User-Visible: no
Issues labelled before the pipeline existed keep their spec straight in dev and
have no issue/NN branch. The publish step quietly exited zero for them, so the
verdict would arrive as a comment and the analysis behind it would be thrown
away — the fifth instance today of a step reporting success by doing nothing.
The document now goes wherever the spec itself lives: the task branch when there
is one, dev otherwise. Publishing also survives dev moving on while the review
ran, which takes up to forty-five minutes, by rebasing once before it gives up.
Four issues are waiting on this — #12, #30, #44 and #52 — each with a spec in dev,
a status label applied during the bulk pass in August and a review that never ran
because nothing was there to raise the event.
Issue: #114
User-Visible: no
The two copies of this file must match byte for byte; a comment line had drifted
by one character. main is the copy the issues event actually reads, so it is the
reference. Trivial in itself, and worth closing anyway: the file's own header
warns that a divergence between these two branches is one of the ways this
pipeline fails quietly.
Issue: #114
User-Visible: no
The guard refused to review any issue the owner had not filed himself. The rule
was meant to keep malformed outside reports out of the pipeline, but it checked at
every step instead of at the entrance, and it duplicated a guarantee the platform
already gives: only someone with write access can apply a label. Applying the
first status label is the owner's explicit decision, and it is the only place the
question belongs.
So the author check is gone. While an issue carries no status label it sits
outside the process and the invariants do not apply; once labelled, the task is in
flight and who filed it stops mattering.
The old rule also cost real work. On #123 an outside bug report had been analysed
and specified before the guard turned it away in nine seconds, and the remedy on
offer was to refile the same thing as the owner's own issue.
Issue: #114
User-Visible: no
The guard refused to review any issue the owner had not filed himself. The rule
was meant to keep malformed outside reports out of the pipeline, but it checked at
every step instead of at the entrance, and it duplicated a guarantee the platform
already gives: only someone with write access can apply a label. Applying the
first status label is the owner's explicit decision, and it is the only place the
question belongs.
So the author check is gone. While an issue carries no status label it sits
outside the process and the invariants do not apply; once labelled, the task is in
flight and who filed it stops mattering.
The old rule also cost real work. On #123 an outside bug report had been analysed
and specified before the guard turned it away in nine seconds, and the remedy on
offer was to refile the same thing as the owner's own issue.
Issue: #114
User-Visible: no
A review label promises work. When the guard declined it wrote the reason to the
run log and nothing else, so the issue sat in a status nobody was acting on and
nobody could tell. #123 showed it: an outside reporter's issue was walked up to
S4-spec-review, the guard refused in nine seconds because only the owner's issues
enter the process, and the issue itself said not a word.
Refusals that a human can act on now become a comment: wrong author, blocked,
review-4. Only when a stage was actually recognised, so an unrelated label change
stays silent.
This is the same defect as the merge conflict that left the label untouched, seen
from the other side. The pattern is worth naming: doing nothing quietly is the
most expensive thing a pipeline can do.
Issue: #114
User-Visible: no
A review label promises work. When the guard declined it wrote the reason to the
run log and nothing else, so the issue sat in a status nobody was acting on and
nobody could tell. #123 showed it: an outside reporter's issue was walked up to
S4-spec-review, the guard refused in nine seconds because only the owner's issues
enter the process, and the issue itself said not a word.
Refusals that a human can act on now become a comment: wrong author, blocked,
review-4. Only when a stage was actually recognised, so an unrelated label change
stays silent.
This is the same defect as the merge conflict that left the label untouched, seen
from the other side. The pattern is worth naming: doing nothing quietly is the
most expensive thing a pipeline can do.
Issue: #114
User-Visible: no
The implementation loop runs typecheck, unit and build. Golden, browser smokes,
performance and the full HA harness run before a beta — after the code review has
passed and the issue already sits in S8-merged. Some defects cannot surface any
earlier, and until now the process had nothing to say about them, so the honest
reading was a second full review cycle at the most expensive possible moment.
The owner's decision: fix it, re-run what failed, and a green run carries the
release on. The gate named the defect precisely and the same gate proves the fix,
so the check is objective and depends on nobody's judgement.
The boundary is written down with it, because "the gate found something" could
otherwise absorb an arbitrary amount of new work. A fix that changes a behaviour
contract, reaches an untouched subsystem or rivals the task in size goes through
the normal flow. Editing a test so it stops failing is concealment rather than
repair — the exception is a defect proven to be in the fixture, as on #89.
The rule also records what it costs: the author judges his own work here, which
the process refuses everywhere else. That is the price of speed at the one point
where a review cycle is dearest, and the compensation is that the re-run command
and its result are written into the issue where the release manager reads them.
Issue: #114
User-Visible: no
A review run always moves the label. The rule is written down because its absence
cost a real stall: a green code review whose merge conflicted left the label alone,
the waiting author polled thirty times and reported the limit as exhausted, and a
verdict that existed reached nobody.
Both documents now say what S6-in-progress means when the verdict was green and
only the merge failed — rebase, not rework, and the verdict still stands. They also
say that a label which did not change means the run failed rather than the work, so
the answer is logs and the owner, not more polling. Cycles are counted per stage.
Issue: #114
User-Visible: no
A green code review whose merge conflicted used to leave the label where it was.
That is a dead end: the author waits for the label to change, so it polled thirty
times and reported the limit as exhausted — on a task the reviewer had already
passed. The verdict existed and nobody could act on it.
The merge step no longer fails the job. It reports whether it merged, and a green
review that did not merge sends the task back to S6-in-progress, because the work
did return to the author — a rebase rather than a code fix, and the comment says
so and says the verdict still stands.
The invariant is now stronger and worth stating plainly: after a review run the
label always changes. A pipeline whose state can stall silently is worse than one
that reports the wrong state loudly.
Issue: #114
User-Visible: no
A green code review whose merge conflicted used to leave the label where it was.
That is a dead end: the author waits for the label to change, so it polled thirty
times and reported the limit as exhausted — on a task the reviewer had already
passed. The verdict existed and nobody could act on it.
The merge step no longer fails the job. It reports whether it merged, and a green
review that did not merge sends the task back to S6-in-progress, because the work
did return to the author — a rebase rather than a code fix, and the comment says
so and says the verdict still stands.
The invariant is now stronger and worth stating plainly: after a review run the
label always changes. A pipeline whose state can stall silently is worse than one
that reports the wrong state loudly.
Issue: #114
User-Visible: no
The step that comments on the issue when a review run dies carried a literal
backslash instead of a line continuation, so gh received four arguments and
--repo ran as a command of its own. The handler for failures would itself have
failed, silently, and only when something had already gone wrong.
bash -n does not catch this: the syntax is valid, the meaning is not. Checking
run blocks now also means looking for a doubled backslash at end of line.
Issue: #114
User-Visible: no
The step that comments on the issue when a review run dies carried a literal
backslash instead of a line continuation, so gh received four arguments and
--repo ran as a command of its own. The handler for failures would itself have
failed, silently, and only when something had already gone wrong.
bash -n does not catch this: the syntax is valid, the meaning is not. Checking
run blocks now also means looking for a doubled backslash at end of line.
Issue: #114
User-Visible: no
The guard counted every verdict comment on the issue, so a spec-review verdict
consumed a cycle from the code-review budget. On #89 the first code review came
out as r2/4. With two spec cycles the second code review would have hit review-4
after a single fix — the limit would have fired on a task nobody had reviewed
twice.
The stage is now resolved first and only its own verdicts are counted, recognised
by the review document named in the comment. If the document is missing the
verdict is not counted: undercounting grants an extra cycle, overcounting would
stop the work early, and of the two mistakes the recoverable one wins.
Issue: #114
User-Visible: no
The guard counted every verdict comment on the issue, so a spec-review verdict
consumed a cycle from the code-review budget. On #89 the first code review came
out as r2/4. With two spec cycles the second code review would have hit review-4
after a single fix — the limit would have fired on a task nobody had reviewed
twice.
The stage is now resolved first and only its own verdicts are counted, recognised
by the review document named in the comment. If the document is missing the
verdict is not counted: undercounting grants an extra cycle, overcounting would
stop the work early, and of the two mistakes the recoverable one wins.
Issue: #114
User-Visible: no
Git skips a hook without the bit and says nothing about it, so the gate
would have reported success by being absent. The API cannot set the mode:
a file pushed that way arrives as 100644.
Issue: #121
User-Visible: no
docs/specs/README.md kept a "Статус ТЗ" column with its own vocabulary — draft,
in implementation, done — next to the labels that already hold the status. Two
dictionaries for one fact drift apart, and these had: the column still called
issues "in implementation" that were closed weeks ago. The table now says only
which issue a spec belongs to.
AGENTS.md was telling agents that PROCESS.md §1 does not cover package.json and
the rest of the configuration, and to report it as missing. It covers them now.
The same paragraph gained the rule that D beats A where paths overlap, which is
what keeps the built bundle under custom_components/houseplan/frontend/ from
reading as product source.
Issue: #119
User-Visible: no
Section 10.1 promised pre-push as the blocking gate that replaces pull requests.
The hook did not exist, so the document promised a check that was not there —
worse than saying nothing, because a promise like that gets relied on. Until now
process-gate ran only as the catch-up job in CI, which reports after the code is
already in dev.
The hook skips branch deletions and tags, and for a branch the remote has not
seen it measures from the merge-base with dev rather than from the root, or every
violation committed before the gate existed would make it impossible to pass. A
missing script does not block a push: old checkouts and worktrees have to stay
usable.
gh is optional on purpose. Reading issue status needs the network, and a hook
that cannot work on a train is a hook people switch off; offline it runs what it
can and CI does the strict pass.
The executable bit is the quiet part. Git skips a hook without +x and says
nothing — the gate reports success by being absent. Measured on a real push:
mode 644 produces zero lines from the gate and the push goes through, 755 stops
it. The API cannot set the bit, so install-hooks restores it on every install.
Issue: #121
User-Visible: no
PROCESS.md 10.2 item 10 asks for this to happen because a beta shipped, not
because someone remembered. The manual cleanup was skipped twice and both times
it broke the invariant that a closed issue carries no status label — the one
thing `verify` leans on. A manual step that falls due right after a successful
release is the worst kind: the work already looks finished, which is precisely
why it gets forgotten.
The job comments the tag, removes the label, then closes. That order is
deliberate: dying between the two steps leaves an open issue without a status,
which is visible and fixable in the flow, where the reverse order would recreate
the breakage this exists to prevent. It ends by asserting that no closed issue
still carries S8-merged — aimed at the defect that actually recurs rather than at
the invariant in general.
The stock token is used on purpose. Events caused by GITHUB_TOKEN do not start
workflows, so stripping the label cannot wake the review pipeline; a PAT here
would turn bookkeeping into a cascade.
Issue: #120
User-Visible: no
Check 2 compared the Issue trailers against whatever branch the working tree
happened to be on, over whatever range it was given. Those two are not the same
set. After a rebase the CI range widens — `before` points at a discarded commit,
the merge-base slides back, and commits that belong to dev arrive carrying other
issue numbers. Every one of them then looks like a violation.
Running the gate over real history from issue/89 with a dev range produced 26
false refusals out of 26 commits, which would have reddened Validate on the next
force-push of any task branch.
The rule now reads origin/dev..HEAD for its own verdict and leaves the event
range to the other checks. A commit that genuinely carries the wrong trailer for
its branch is still caught; the integration test covers both directions.
Issue: #105
User-Visible: no
The canon moved into the repository in #112 and then stood still while the
process kept moving. A document that lags is worse than no document: an agent
reading it as truth acts on rules that no longer exist. It promised a pre-push
hook that was never written, named labels in Russian that the repository has
never used, listed a status set the gate no longer applies, and said nothing at
all about the event-driven pipeline — the largest mechanism the process has.
Label names are now English throughout and S8-merged is documented. Section 1
covers the configuration files the gate kept reporting as unclassified, and
records that D beats A where paths overlap, since the built bundle lives inside
custom_components/houseplan/frontend/. Section 10.1 admits that pre-push does
not exist. Section 10.2 matches ALLOWED_STATUS in scripts/process-gate.mjs,
including the two caveats that only surfaced once the pipeline ran. Section 10.4
is new and documents the four silent-failure traps that cost a working day each.
The source-of-truth order now says that actual automation outranks its own
description — this document included.
Issue: #119
User-Visible: no
Practice had already diverged from the documents: #105, #112, #114 and #116 were
all done without a spec and without review, and that was right. Nothing said it
was allowed.
The test is mechanical — not a single class A file — rather than left to the
executor's judgement, because a loose reading is exactly how product changes
would learn to skip review.
Issue: #118
User-Visible: no
PROCESS.md 10.2 describes scripts/process-gate.mjs; the script never existed.
Commits go straight to dev without PRs and GitHub blocks nothing on its side, so
until now the only thing standing between the process and rule #1 was the good
faith of whoever was committing. Hooks catch a violation on the author's machine
but --no-verify walks past them; this job is the catch-up pass that cannot be
skipped locally.
Checks 1-7 offline, 8 through gh, plus the escalation of check 3: a class A
commit with neither a spec file nor the `small` label is a failure, not a
warning. Check 8 is fail closed — an unreachable or closed issue is a refusal,
never a silent pass.
Two things surfaced while wiring it up and are recorded in the script header.
S8-merged had to join the allowed statuses: the pipeline merges into dev before
it moves the label, so Validate reads the issue already advanced and a strict set
would redden every accepted task. And the status question now applies only to
class A/B commits — asking it of a review document would fail every time, since
that document lands while the issue sits in S4-spec-review or S7-code-review.
Issue: #105
User-Visible: no
Review fires from the label and runs on its own, but nothing was picking the
result up: the author reported "handed over for review" and stopped, so the
conveyor stalled until the owner said a sentence. The author now polls the label
and continues from whatever it became.
Also drops the merge-into-dev standing permission: the pipeline does the merge
before setting S8-merged, so a hand merge would race it.
Issue: #114
User-Visible: no
The label asserts the code is in dev. The workflow used to set it on a green
code review while the commits were still only on the task branch, so between
the verdict and the author's merge the state machine stated something untrue —
which is exactly what happened on #104.
The merge now runs inside the pipeline, before the label. A conflict leaves
the issue in S7-code-review and comments instead.
Issue: #114
User-Visible: no
Keeps dev identical to main so the broken revision does not come back at the
next promotion. The workflow only fires from the default branch, but a stale
copy here would overwrite the working one.
Issue: #114
User-Visible: no
PROCESS.md wants a review document in docs/reviews/; the CI reviewer could
only leave a comment, and flagged the gap itself. It may now write there.
What lands in the commit is decided by the workflow, not by the model: every
path outside docs/reviews/ is reverted before staging, and the commit carries
the usual trailers so the provenance gate accepts it.
Issue: #114
User-Visible: no
The multi-line --body started at column zero, which ends the YAML block
scalar. The parser silently truncated the run script and left an unclosed
double quote, so the whole workflow became unusable and blocked the code
review on #104.
The body now goes through a heredoc. Validating YAML alone did not catch
this; every run block is checked with bash -n from now on.
Issue: #114
User-Visible: no
Pushing the hook through the GitHub contents API dropped its mode to 100644.
assertHookMode caught it on the next commit, which is the gate working as
intended — a non-executable commit-msg would simply never run.
Also brings dev in line with the turn-limit fix already on main.
Issue: #116
User-Visible: no
The r2 spec review on #104 produced a complete green verdict and then failed
on --max-turns 40 at turn 43, so the label step never ran and the transition
had to be reconciled by hand. Forty was a guess; a review that reads SCOPE,
AGENTS, PROCESS, the issue thread and the spec exceeds it routinely, and a
code review that also runs gates needs far more.
The real guard against a runaway run is the job timeout, not the turn count.
Issue: #114
User-Visible: no
Agents were escalating technical calls. The owner answers what a person sees
or does and how much user-visible change belongs in an issue; storage, module
layout, naming, test strategy and migration mechanics are settled by the
agents, recorded as assumptions and challenged in review.
Issue: #114
User-Visible: no
Without it S8-merged would be a lie: the label asserts the code is in dev,
while the branch-only push leaves it on the task branch. What lands is what
the reviewer just accepted, and dev is allowed to carry unreviewed code
anyway, so the merge adds no risk the branch did not already carry.
Issue: #114
User-Visible: no
The hook parsed the message path with basename, an external command. Where
it is missing from PATH the substitution yields an empty string, set -e does
not trip on it, and the MERGE_MSG guard silently stops working — a generated
merge commit would then be rejected for missing trailers it cannot have.
POSIX parameter expansion needs no external command and behaves the same in
sh, dash, bash and Git Bash.
Issue: #116
User-Visible: no
The reviewer runs in CI and can only read the remote, so an unpushed spec or
commit either stalls the review or points it at the wrong tree. Pushing
issue/<NN>-slug now needs no command; dev, main, merges, tags and releases
still do.
Issue: #114
User-Visible: no
The first live run failed with "Could not fetch an OIDC token": the action
needs id-token: write to authenticate the GitHub App.
The reviewer also checked out dev, where the material under review does not
exist yet — specs and code are committed to issue/<NN>-slug. The job now
switches to that branch when it is pushed, and warns loudly when it is not.
Issue: #114
User-Visible: no
The first live run failed with "Could not fetch an OIDC token": the action
needs id-token: write to authenticate the GitHub App.
The reviewer also checked out dev, where the material under review does not
exist yet — specs and code are committed to issue/<NN>-slug. The job now
switches to that branch when it is pushed, and warns loudly when it is not.
Issue: #114
User-Visible: no
Adds .github/workflows/process.yml. A status label change is the trigger:
S4-spec-review runs the spec review, S7-code-review runs the code review,
and the verdict decides the next label. Only a green verdict advances;
yellow and red return the task to its author. Cycle limits (4, or 2 on the
light track) are counted from the verdicts already posted on the issue.
Labels are moved with HP_PROCESS_TOKEN, not GITHUB_TOKEN, so the change
emits an event and the chain continues.
Issue: #114
User-Visible: no
Adds .github/workflows/process.yml. A status label change is the trigger:
S4-spec-review runs the spec review, S7-code-review runs the code review,
and the verdict decides the next label. Only a green verdict advances;
yellow and red return the task to its author. Cycle limits (4, or 2 on the
light track) are counted from the verdicts already posted on the issue.
Labels are moved with HP_PROCESS_TOKEN, not GITHUB_TOKEN, so the change
emits an event and the chain continues.
Issue: #114
User-Visible: no
Publishes the 538-line process canon into the repository, replacing the
51-line provenance stub that pointed at a non-existent .agents/PROTOCOL.md.
Rewrites AGENTS.md: product context first, labels as the canonical status,
rule #1 with the status check, change classes, trailers, push cadence,
Codex/Claude roles and review cycle limits.
Issue: #112
User-Visible: no
The HACS submission check does not read hacs.json to find the integration: it
globs `*manifest.json` over the whole clone of the default branch and exits 1
unless there is exactly one (hacs/default, scripts/helpers/integration_path.py).
Three files matched — the two stand-only integrations added on 2026-07-31 and
the golden baseline index added on 2026-08-11 — so the Hassfest job of PR #9004
went red five weeks into the review queue, with a log that named no file.
The stand manifests ship as manifest.template.json and demo/stand/install.sh
renames them at install time; the golden index becomes baselines-index.json
(the exported constant keeps its name, so no consumer changes).
test/repo-hygiene.test.mjs fails if a second manifest ever appears, and the
existing golden-policy assertion — which compared against 'manifest.json' and
happily passed on 'baseline-manifest.json' — now checks the suffix.
HACS will install from the named asset instead of the auto zipball, so
GitHub's public download counter becomes a per-version install metric.
The workflow also has a manual dispatch to backfill an existing release.
Метка \`$LABEL\` обещает работу, которая не начнётся, поэтому статус лучше вернуть в предыдущий — иначе задача простоит здесь бесконечно. [Прогон](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }})."
stage=""
}
# Автор issue здесь не проверяется (решение владельца 2026-08-13).
# Проверка стоит на входе в процесс, а не на каждом шаге: как только
# задача получила статусную метку, она в работе, и кто её завёл — не
# имеет значения. Само присвоение метки и есть явное подтверждение
# владельца, причём проверенное платформой: метки может ставить только
# тот, у кого есть право записи в репозиторий. Прежняя проверка здесь
# дублировала эту гарантию и заставляла переоформлять чужие отчёты
# своими issue — чистая работа впустую, как на #123.
if [ -z "$stage" ]; then
:
elif [ "$BLOCKED" = "true" ]; then
refuse "стоит blocked — конвейер не запускается" \
"на issue стоит \`blocked\` — задача ждёт внешнего решения. Снять метку, когда решение принято."
elif [ "$EXHAUSTED" = "true" ]; then
# Метку снимает владелец, а не конвейер: автоматика, отменяющая
# остановку работы, дороже ручного снятия. Но пересчёт печатается —
# метка могла остаться от прежнего правила, когда бюджет тратил и
# зелёный вердикт (#227).
stale=""
if [ "$spent" -lt "$limit" ]; then
stale=" Пересчёт по действующему правилу: блокирующих циклов $spent из $limit — метка могла остаться от прежнего правила, когда бюджет тратил любой вердикт. Снять её может владелец."
fi
refuse "стоит review-4 — решение за владельцем" \
"на issue стоит \`review-4\`: лимит циклов ревью исчерпан, дальше решает владелец — разделить задачу, отклонить или арбитраж (PROCESS.md §4).$stale"
"Лимит циклов ревью исчерпан: блокирующих циклов $spent из $limit на этапе \`$stage\` (заход $attempt). Следующего захода нет: решение владельца — разделить задачу, отклонить или арбитраж (PROCESS.md §4).
Учтены вердикты с блокирующими находками — зелёные бюджет не тратят:
$spent_list"
stage=""
else
echo "этап $stage, заход $attempt, блокирующих циклов $spent из $limit"
fi
echo "stage=$stage" >> "$GITHUB_OUTPUT"
echo "cycle=$attempt" >> "$GITHUB_OUTPUT"
echo "spent=$spent" >> "$GITHUB_OUTPUT"
echo "limit=$limit" >> "$GITHUB_OUTPUT"
review:
needs:guard
if:needs.guard.outputs.stage != ''
runs-on:ubuntu-latest
# Время — единственный настоящий ограничитель зациклившегося прогона.
timeout-minutes:45
steps:
- uses:actions/checkout@v7
with:
fetch-depth:0
ref:dev
# Иначе в конфиге git остаётся креденшел GITHUB_TOKEN, и push с
# мёртвым PAT молча уходит от github-actions[bot] — 403 при
# contents: read. Отказ обязан быть громким и правильным.
persist-credentials:false
# Живость PAT проверяется ДО ревью. На #150 истёкший токен обнаружился
# только на публикации документа — после сорока минут работы ревьюера.
- name:Секрет HP_PROCESS_TOKEN жив
env:
GH_TOKEN:${{ secrets.HP_PROCESS_TOKEN }}
run:|
if [ -z "$GH_TOKEN" ]; then
echo "::error::HP_PROCESS_TOKEN пуст — секрет удалён или недоступен"
exit 1
fi
if ! login=$(gh api user -q .login 2>/dev/null); then
echo "::error::HP_PROCESS_TOKEN не аутентифицируется — истёк или отозван. Обновить: Settings -> Secrets and variables -> Actions -> HP_PROCESS_TOKEN"
exit 1
fi
echo "токен жив, действует от: $login"
# Окружение готовит workflow, а не модель своими ходами. Раньше промпт
# велел ревьюеру самому выполнить `npm ci`: минуты уходили на установку без
# кэша, платились из бюджета 45 минут и из лимитов подписки, а ходы модели
# тратились на работу инфраструктуры. В validate.yml кэш стоит на всех
# тяжёлых job, здесь его не было.
- uses:actions/setup-node@v7
with:
node-version:22
cache:npm
# Материал ревью живёт в ветке задачи: ТЗ в docs/specs/ и код коммитятся
# в issue/<NN>-slug. Если ветка запушена — переключаемся на неё, иначе
# ревьюер прочтёт dev и не найдёт того, что должен оценивать.
- name:Перейти на ветку задачи
id:branch
env:
NUM:${{ github.event.issue.number }}
run:|
# Свежая по последнему коммиту, а не первая по алфавиту: на #150 рядом
# жили ветка ТЗ и ветка реализации, и head -1 выбрал устаревшую.
echo "note=Ветка приведена к dev конвейером до ревью: поверх легло $behind коммит(ов) dev, $short_before -> $short_after. После ребейза это другой код (§7.2) — разбор полный, а не по дельте." >> "$GITHUB_OUTPUT"
echo "ветка $BRANCH приведена к dev: $short_before -> $short_after"
# Конфликт возвращает задачу автору ДО ревью. Инвариант «после прогона
# метка меняется всегда» при этом держится: возврат в S6-in-progress —
# тоже смена метки, и автор не ждёт впустую.
- name:Конфликт с dev — вернуть автору без ревью
if:steps.rebase.outputs.conflict == 'true'
env:
GH_TOKEN:${{ secrets.HP_PROCESS_TOKEN }}
NUM:${{ github.event.issue.number }}
BRANCH:${{ steps.branch.outputs.name }}
run:|
cat > /tmp/stale.md <<EOF
**Ревью не запускалось:** ветка \`$BRANCH\` не ребейзится на \`dev\` без конфликта. Код никто не читал, вердикта нет, цикл ревью не израсходован.
Проверка стоит до ревью намеренно: конфликт всё равно вернул бы задачу, но уже после сорока минут работы ревьюера и потраченных лимитов.
Задача переведена в \`S6-in-progress\`. Осталось:
1. \`git fetch origin\`, затем \`git rebase origin/dev\` в ветке задачи, разрешить конфликт;
echo "::warning::ветка $BRANCH не сливается в dev без конфликта"
cat > /tmp/conflict.md <<EOF
**Код-ревью зелёное — вердикт выше в силе, переделывать работу не нужно.** Не удалось только слияние: ветка \`$BRANCH\` конфликтует с \`dev\`.
Задача переведена в \`S6-in-progress\`, потому что работа вернулась к автору. Осталась не правка кода, а ребейз:
1. \`git fetch origin\`, затем \`git rebase origin/dev\` в ветке задачи, разрешить конфликт;
2. запушить ветку;
3. вернуть метку \`S7-code-review\`.
Повторный прогон ревью — не формальность: после ребейза на новый \`dev\` это другой код, и принимать его без проверки нельзя. Цикл считается по этапу, лимит на код-ревью тратится отдельно от ревью ТЗ.
# Тело через heredoc, а не многострочный --body: строка с нулевым
# отступом обрывает блок YAML и оставляет незакрытую кавычку.
cat > /tmp/failure.md <<EOF
Автоматическое ревью не отработало: [прогон]($RUN_URL). Статусная метка не менялась, задача осталась на месте.
Если вердикт выше всё же опубликован — сбой произошёл после него. Перестановку метки в этом случае выполняет чат обслуживания или владелец, но не автор задачи: автор не толкует вердикт о своей же работе.
@@ -6,22 +6,424 @@ House Plan is one HACS package with two parts plus a demo harness:
- **Storage integration** (`custom_components/houseplan/`, Python) — the Home Assistant backend.
- **Demo harness** (`demo/`) — a self-contained Playwright page (`demo/srv/demo.html`) that renders the card against a fake `hass`, used for screenshots and the `smoke_*.mjs` end-to-end suite.
Standard commands live in `package.json` scripts, `CONTRIBUTING.md`, and `docs/DEVELOPMENT.md`. Read `docs/ARCHITECTURE.md` and `docs/STATUS.md` before non-trivial changes.
## Read this first
## Cursor Cloud specific instructions
**`docs/SCOPE.md` before anything else.** It was fixed with the owner and states
its own authority: features are built, improved and accepted **only** if they
serve a job listed there. It carries the mission, the three personas, the core
user jobs and the out-of-scope list.
The startup update script already runs `npm ci`, provisions a Python 3.13 backend venv at `.venv-backend`, and installs Playwright Chromium. You do not need to reinstall dependencies.
Its central consequence: **View mode is the product for two of the three
personas.** Editors are admin-only tools and must never leak interactions into
View.
- **Frontend** (from repo root): `npm run typecheck`, `npm test` (node:test, ~270 tests), `npm run build`. After building, keep the integration copy in sync — `cp dist/houseplan-card.js custom_components/houseplan/frontend/`. CI enforces `cmp dist/houseplan-card.js custom_components/houseplan/frontend/houseplan-card.js` byte-for-byte.
- **Backend HA-harness tests need Python 3.13, not the system 3.12.** Run them with the venv: `.venv-backend/bin/python -m pytest tests_backend/ -q` (126 tests). Running `python3 -m pytest tests_backend` on the system 3.12 silently **skips** the `test_ha_*.py` harness tests (`conftest.py` ignores them when `homeassistant` is not importable) and runs only the ~83 pure tests.
- **Running the app / smoke suite**: build a fresh bundle and copy it into the demo assets first — `npm run build && cp dist/houseplan-card.js demo/srv/assets/houseplan-card.js` — then run `node demo/smoke_*.mjs`. The committed `demo/srv/assets/houseplan-card.js` is a stale snapshot; testing it reports green about code that no longer exists. No real Home Assistant server is required: `demo/srv/demo.html` stubs `hass`, registries and `callService`.
- **Demo harness render quirk**: the fake `hass` in `demo.html` is set once, so opening the page directly in a browser renders the floor plan but **device icons only appear after a re-render** (an F5 refresh, or nudging `card.hass = {...card.hass}`). The smoke launcher `demo/serve.mjs` already does this nudge; a plain browser session does not. This is a harness limitation, not a card bug.
- **Known environment-sensitive smoke**: `demo/smoke_opening_measure.mjs` fails two sub-checks (`place_dialog_x_magnetised`, `place_committed_x_center`) under the pinned Chromium — a `1e-6`-tolerance magnet-snap on the opening-*placement* path. It reproduces against the pristine committed bundle, so treat it as pre-existing/pixel-precision, not a regression you introduced.
- **Owner's local Windows checkout is invisible here.** Path
`C:\Users\Sergey\Downloads\dev\houseplan-dev` (workflow notes + often
unpushed edits) is **not mounted** into managed Cloud Agent VMs. Do not
expect to `ls` or diff that folder. To bring local work into the cloud
agent: push a branch to GitHub and say its name, or run a **local** Cursor
Agent / My Machines worker inside that checkout. Day-to-day source of truth
for cloud sessions remains **`origin/dev`** (minors) and **`origin/main`**
(releases) — see `docs/STATUS.md`.
For work that changes visible behaviour, also read `docs/USER-GUIDE.ru.md` —
interface wording comes from there and is not invented, or the UI starts speaking
developer.
Then `PROCESS.md` (the full process), `docs/STATUS.md` (where the release line
is), and for non-trivial changes `docs/ARCHITECTURE.md` plus the canonical
document of the subsystem you touch: `SUN.md`, `LIGHT.md`, `CANVAS.md`,
| **B — gates and tooling** | `test/**`, `tests_backend/**`, `demo/**`, `scripts/**`, `.github/workflows/**`, `rollup.config.mjs`, `tsconfig*.json` | yes; may reuse the issue it covers |
| **C — documentation** | `docs/**`, `README*`, `CHANGELOG*`, `AGENTS.md` | not if it is part of its issue's DoD |
| **D — generated** | `dist/**`, `custom_components/houseplan/frontend/**`, `demo/golden/baselines/**` | never changes on its own. The stand copy `demo/srv/assets/houseplan-card.js` is no longer committed (#255): build it with `npm run bundle:sync` |
The table above is a summary; `PROCESS.md` §1 is the authority and now covers the
configuration files this one omits — `package.json`, `package-lock.json`,
`pytest.ini`, `.gitignore`, `.gitattributes`, `.githooks/**` and the rest of
`.github/**` are class B. Where paths overlap, **D beats A**: the built bundle
lives inside `custom_components/houseplan/frontend/` and would otherwise read as
Never invent a review link and never rewrite published history to satisfy
trailers. `.githooks/commit-msg` and the `provenance` CI job both run
`scripts/validate-commit-provenance.mjs`.
Branch: `issue/<NN>-slug`. Direct commits to `dev`, no PR — the owner's decision;
CI checks after the fact, and a violation is fixed with a follow-up commit, never
a force-push.
**Push after every task, not before a beta.** While work sits unpushed there is
nothing to review, and reviewing twenty tasks at once is not review. `dev` may hold
unreviewed code while a task is in flight; what matters is its state when the
reviewer says it is accepted.
**Standing permission: push `issue/<NN>-slug` without asking.** The reviewer runs
in CI and can only read what is on the remote — an unpushed spec or commit means
the review either stalls or judges the wrong tree. Pushing a task branch publishes
nothing to users and does not touch the integration branch, so it needs no command.
**Do not merge into `dev` by hand.** On a green code review the pipeline rebases
the task branch onto `dev`, pushes it, and only then sets `S8-merged` — the label
asserts the code is in `dev`, so the merge has to happen first or the label lies
in between.
If the rebase conflicts the pipeline says so in the issue and sends the task back
to `S6-in-progress`. The verdict still stands: nothing needs reviewing again, the
remaining work is the rebase. Resolve it, push the branch, re-apply
`S7-code-review`. The second review run is not a formality — after a rebase onto a
moved `dev` this is different code, and accepting it unchecked is how regressions
arrive. Cycles are counted per stage, so a code review spends its own budget.
Everything else still requires the owner's explicit command: pushing `main`,
creating tags, publishing betas and releases, closing issues.
## Working trees (#115)
One checkout, one `HEAD`: two agents sharing a directory inherit each other's
branch, and twice in one hour a commit landed on someone else's task branch that
way. The layout is therefore fixed:
- **`houseplan-card-src/houseplan-card`** — the author's tree. Task branches live
here; nobody else commits in it. Unfamiliar local changes belong to the author
or the owner — never reset or clean them away.
- **`houseplan-card-src/hp-dev`** — the owner's worktree, permanently on `dev`. For owner-side operations that must not disturb the
author's tree: pushing `dev`, restoring a hook's executable bit, emergencies.
- **The reviewer and the infrastructure agent own no local tree.** The reviewer
runs in CI on a fresh checkout. The infrastructure agent reads via `git show`
and publishes through the GitHub API; it makes no local commits at all, so it
needs no `HEAD` of its own. Its scratch worktrees live outside the repo and are
pruned after use.
A worktree is only usable on the machine that created it: the `.git` file records
an absolute path in that machine's format. One created from a Linux sandbox is
dead on Windows and vice versa — create worktrees on the machine that will use
them, which for `hp-dev` means the owner's.
## Two-agent workflow
**Codex** writes analysis, specs and all product code. **Claude** reviews specs and
code and owns infrastructure and distribution. The owner rules on disputes, closes
issues and commands releases.
Author and reviewer are different models, which is what "a fresh session without
implementation context" means in practice. The reviewer never edits product code;
the author never grades their own work.
**Infrastructure-only work runs outside this flow.** CI, scripts, labels, demo
stands, the landing page and distribution are Claude's alone, and running them
through spec-writing and review buys nothing: the spec would restate what is
already unambiguous, and author and reviewer would be the same role. So no spec
file, no spec review, no code review, no walk through `S1`…`S8`.
The test for "infrastructure only" is mechanical: **not a single class A file** —
nothing under `src/**`, no `custom_components/**/*.py`, no manifests, no i18n. A
task that touches class A even once is not infrastructure and takes the full flow;
there is no such thing as "mostly infrastructure". The strictness is deliberate:
a loose reading would turn this into the route by which product changes skip
review.
What stays mandatory either way: an issue exists, both trailers are on every
commit, `typecheck`, `test` and `build` are green, and any non-obvious decision is
written down in the code or the issue rather than kept in someone's head.
**Review starts by itself.** Applying `S4-spec-review` or `S7-code-review` fires the
pipeline, which reviews without anyone asking and takes ten to forty-five minutes.
**Having applied one of those labels, wait for the result instead of ending the
session.** Reporting "handed over for review" stops a conveyor that could have kept
moving on its own. An agent has no clock — it exists only during its own turn — so
waiting means polling: every 90 seconds, at most 30 times. A single long sleep hits
the command timeout. Watch the **label**, not the comment: the label is the state,
the comment only explains it. Do not wait at all while `blocked` is set — the task
is waiting on the owner, not on the reviewer. On exhausting the attempts, stop and
tell the owner: a failed run leaves the label where it was, forever.
What the new label means:
| Now reads | What happened | What you do |
|---|---|---|
| `S5-ready` | the spec is accepted | write the code |
| `S3-spec` | the spec came back | read the verdict, revise, re-apply `S4-spec-review` |
| `S6-in-progress` | the code came back | revise, re-apply `S7-code-review` — **or**, if the verdict was green and only the merge conflicted, just rebase and re-apply. The comment says which |
| `S8-merged` | accepted and already in `dev` | nothing |
| `review-4` | the cycle limit is spent | stop, the owner decides |
**After a review run the label always changes.** If it did not, the run itself
failed rather than the work — say so to the owner instead of polling on.
**A failed pre-release gate does not send the issue back to review.** The
implementation loop runs only typecheck, unit and build; golden, browser smokes,
performance and the full HA harness run before a beta, which is after the code
review has passed and the issue sits in `S8-merged`. Some defects cannot surface
any earlier.
Fix it, re-run what failed, and a green run is enough for the release to continue.
The issue stays in `S8-merged`. Record the **exact command and its result** in the
issue — "verified" without a command proves nothing. Trailers as usual, and
`User-Visible: yes` still means both changelogs in the same commit.
The exception covers repairing the defect the gate named, not carrying on
development under the name of a repair. It goes through the normal flow — a new
issue, or back to `S6-in-progress` — if the fix changes a behaviour contract, gives
the user something new, reaches a subsystem the task never touched, or is
comparable in size to the task itself. And editing the gate so it stops failing is
concealment, not repair; the exception is a defect proven to be **in the fixture**,
as on #89, where the sun sat at azimuth 180° and the only window faced north, so no
ray was ever built.
Baselines are still accepted only via `npm run golden:accept -- --reviewed` on a
complete Linux CI artefact. "So the gate goes green" is not a reason.
The exchange happens in **issue comments** — there is no local message bus. Verdict
format:
```text
Verdict: green/yellow/red · cycle r<N>/4 · High: N · Medium: N → in-task | #… · Document: …
```
High blocks. A Medium finding INSIDE the task's scope is fixed within the task:
with no High findings the verdict is yellow, the author fixes it and the fix
passes another review cycle — no separate issue (owner's decision 2026-08-19,
#202: filing and servicing an issue costs far more than fixing in place). Only
a Medium finding OUTSIDE the scope becomes its own issue — foreign scope is
never patched from this branch. Low is fixed or waived with a note
in the review document. A yellow verdict is legitimate even when every acceptance
criterion passes, if the change does not solve the stated scenario or degrades a
neighbouring one.
**Four review cycles** (two on the light track). The counter lives in the document
name, `-r1`…`-r4`; the fourth adds the `review-4` label. There is no fifth attempt:
the owner splits the task, rejects it, or arbitrates.
On the light track (`small`: complexity ≤3, one surface, no config migration, no
new UX contract, no perf or touch impact — all at once) the spec lives in the issue
body and the spec review is a comment. Code review is never skipped.
## Specs
`docs/specs/<NN>-<slug>.md`, linked to its issue in both directions. Required
sections are in `PROCESS.md` §7.1, plus two product ones: which persona meets this,
on which surface, at what moment; and what the person sees before and after, in one
sentence without implementation terms.
**Ambiguity is asked, not guessed — but only product ambiguity.** A guess written as
fact is the worst kind of defect: it passes review because it looks like a decision.
The owner answers exactly two kinds of question: **what a person sees or does**, and
**how much user-visible change belongs in this issue**. Behaviour in a boundary case,
which persona wins when two conflict, what counts as acceptable degradation, whether
a neighbouring behaviour is in scope here or becomes its own issue.
Everything a user cannot observe is yours to settle: where state is stored, which
module carries the guard, naming, file layout, test strategy, migration mechanics,
development policy. Decide it, record it in an explicit "assumed, change freely"
block, and let the reviewer challenge it. A technical disagreement between author and
reviewer is settled by the verdict, not by the owner; it reaches him only when the
cycle limit is exhausted.
Split a mixed question instead of escalating all of it. "Where does this state live"
is technical. "Does it survive a page reload and follow the plan across screens" is
product. Ask the second, decide the first.
Ask in one batched issue comment, each question carrying a proposed default, and put
`blocked` on top of `S3-spec` while waiting. A question with a default costs the
owner seconds; one without costs him minutes.
## Gates
```
npm run typecheck
npm test
npm run build
npm run inventory # the only correct way to get test counts
```
Never copy test counts into documents by hand; they go stale in days.
After building, keep all three bundle snapshots in sync — CI compares them
virtual walls and a visual decor layer, all drawn with clicks; room resize
by dragging walls, with live lengths and areas as you drag; smart
alignment guides and a live ruler in real meters/feet.
- 🖼 **A backdrop you can move and scale** — drag the floor-plan picture into
place and pull a corner to size it, with its real size in metres shown as
you drag, so the drawing and the photo of your plan finally line up.
- 💡 **Lights toggle on click** out of the box; wall-switch markers can control
whole groups of lights (works for dumb switches and stateless remotes too).
- 🌒 **“Light sources” fill** — a dark house where every lit lamp casts a pool
of its own color that spills through doorways and open zone boundaries.
- ☀️ **The sun on the plan** — set the compass and the backdrop lives with
the day (white noon → golden hour → deep night), while windows on exterior
walls cast real wedges of sunlight into the rooms; optional cloud cover
from a weather entity.
- 🪟 **Curtains and blinds open on a tap** — one action opens, closes or
stops a cover, and the icon itself morphs between open and closed while a
soft ring pulses as it travels.
- 🌡 **Room cards** with temperature, humidity, Zigbee LQI and light count;
comfort-range temperature fills, per-room signal heatmap.
- 🚪 **Doors, windows and locks** with contact sensors — unlocking is always an
explicit button, never an accidental tap.
- 📺 **Kiosk mode** for wall tablets and TVs: fullscreen, swipe between floors,
auto-carousel, per-screen icon sizes.
- 🤖 **Live robot vacuums** — the dock marker stays put while a round puck
drives the plan in real time, pouring its path out from under itself;
current and previous cleanup runs are recorded server-side. Calibration is
one click (rooms matched by name) or a drag-and-stretch overlay. Works with
Xiaomi Cloud Map Extractor, Tasshack dreame-vacuum and Valetudo.
- 🔔 New devices appear automatically with a red “new” dot; the layout is stored
**server-side** — one shared plan for every user and screen, synced live.
Setup is entirely graphical: no floor-plan YAML, Inkscape, or external editor.
Plan data and device positions live on the Home Assistant server and stay in
sync across screens.
---
> **Edit on a desktop computer.** View and kiosk are fully supported on phones
> and tablets. The editors are designed primarily for a mouse and keyboard;
> individual touch editing operations may be awkward or unavailable. See the
> exact [touch support contract](docs/TOUCH-SUPPORT.md).
## What it is and why
<!-- docs-section: features -->
House Plan shows your smart home the way it actually looks — on a floor plan. Instead of long lists of entities, you see rooms and devices in their real places: where the leak is, what the temperature is in the kids' room, whether the light is on in the hallway, whether the gate is open.
## What House Plan provides
This is convenient when:
- **Live state and safe actions.** Lights and other safe devices can toggle from
the plan; a lock cannot be opened by an accidental plan tap.
- **Three built-in editors.** Plan creates rooms, walls and openings; Device
places and configures markers; Background adds lines, labels and furniture.
- **Area-aware rooms.** New devices appear automatically, while room cards can
show temperature, humidity, light state and average LQI.
- **Light and environment.** Room fills, lamp Glow, wall shadows, a day-cycle
backdrop and sunlight through windows.
- **Doors, windows, gates and vacuums.** Openings follow real contacts and locks;
a robot can show its position, dock and travelled path.
- **Several floors and screens.** Space tabs, swipe navigation, local viewport,
and a separate initial floor for each card.
- **Wall-display kiosk.** A plan-only view with fullscreen navigation and icon
sizes saved for that display.
- you have many devices and lists are awkward to use;
- you need to grasp the state of the house "at a glance";
- you want to give access to family members — anyone can figure out a picture;
- you want a beautiful overview screen for a wall-mounted tablet.

The integration consists of two parts that are installed together:
<!-- docs-section: first-run -->
- **the Lovelace card** `houseplan-card` — the interactive plan itself;
- **the server-side component** — stores the room markup and icon positions in Home Assistant, so the plan is identical in all browsers and on all devices.
## Your first working room
---
1. Install the integration and add the card to a dashboard.
2. Create the first **space**: upload SVG/PNG/JPG/WebP, reuse an uploaded image,
or choose no image and draw the plan by hand.
3. In Plan, select **Room outline**, place vertices, and click the first point to
close the outline.
4. Name the room and bind it to a Home Assistant area. Use “No area” for a room
that has no devices.
5. Open Device: devices from the bound area are already placed; drag their
markers to the correct positions.
6. Optionally use Background for lines, text and furniture.
7. Return to View. The plan now displays live state and accepts safe actions.
## How it differs from alternatives

A house plan in Home Assistant is usually built with `picture-elements`,
`ha-floorplan`, or newer GUI cards that draw walls and furniture in the
dashboard. Those either lock you into YAML/SVG, or store the plan in the
Lovelace card config. House Plan is a **shared live map** backed by a Home
Assistant integration:

| | House Plan | picture-elements / ha-floorplan | GUI draw cards (e.g. easy-floorplan) |
|---|---|---|---|
| **Setup** | Entirely through the UI, with the mouse | Manual YAML / Inkscape SVG | In-card drawing of walls & furniture |
| **Adding devices** | Automatic, by HA **area** | You type every entity by hand | Place entities by hand on the drawing |
| **Icon coordinates** | Drag with the mouse | Count pixels into YAML | Drag on the canvas |
| **Storage** | On the HA server (`.storage`, shared, multi-client) | In the dashboard YAML | In the card / dashboard YAML |
| **Overlays** | Glow, climate, LQI, sun, vacuums, kiosk | Whatever you script in SVG/CSS | Varies by card |
| **Zoom** | Smooth vector zoom | Usually a fixed image | SVG / virtual canvas |

**One sentence:** House Plan is the shared, area-aware live map of your home —
not a drawing app for furniture. If you want to sketch sofas and walls from
scratch, a draw-centric card may fit better; if you already have a plan image
and a real HA areas registry, House Plan keeps every tablet on the same layout.

Key advantages in short:

- **No code at all.** Everything — spaces, rooms, devices — is configured with clicks.
- **Automatic device placement.** Outline a room and bind it to a Home Assistant area — the devices of that area appear on the plan by themselves.
- **Manual additions of your own.** Any device, group or even a "virtual" point can be placed on the plan manually, with a name, icon, model, link and an attached PDF manual.
- **Live states.** Temperature, Zigbee signal strength, on/off, open/closed — everything updates in real time.
Icon colors follow one principle — **yellow means the device is doing its main job right now**:
a light is shining, a socket is powering, a fan is spinning, media is playing, a vacuum is
cleaning, a radiator valve is actually heating (not merely enabled). Orange = open / unlocked.
A pulsing red ring = an emergency (leak, smoke, gas). An RGB bulb's colour lives in its glow
spot (glow fill), where the spot itself is the on/off indicator and the badge stays standard.
A translucent icon = unavailable. Dark = idle.
- **Crisp zoom.** Zooming in does not "blur" the picture: the plan, labels and icons remain vector-sharp at any scale.
Every workflow and edge case is in the [full user guide](docs/USER-GUIDE.md).
The [Background editor contract](docs/DECOR-EDITOR.md) and
[vacuum guide](docs/VACUUM.md) are the authorities for those subsystems.
---
## Wall tablet / TV (kiosk mode)
Add the card to a dedicated dashboard with a **panel view** and set `kiosk: true`
(or tick "Wall device (kiosk) mode" in the card editor):
```yaml
type:custom:houseplan-card
kiosk:true
cycle:0# seconds between auto space switches, 0 = off (nice for TVs)
```
No header, no editors — just the live plan. Swipe to change floors (at 1:1),
pinch to zoom, double-tap to reset. Long-press an empty spot for 3 seconds to
tune icon and text sizes for THIS screen (saved per device). To hide Home
Assistant's own header use the companion app's kiosk settings or the
[](https://my.home-assistant.io/redirect/hacs_repository/?owner=Matysh&repository=houseplan-card&category=integration)
[](https://my.home-assistant.io/redirect/hacs_repository/?owner=Matysh&repository=houseplan-card&category=integration)
1. In HACS open **⋮ → Custom repositories**.
2. Add `https://github.com/Matysh/houseplan-card` as an **Integration**.
3. Install House Plan and restart Home Assistant.
4. Open **Settings → Devices & services → Add integration → House Plan**.
### Via HACS (recommended)
The card is registered automatically. If you manage Lovelace resources
manually, use the URL served by the integration:
1. Open **HACS → menu (⋮) → Custom repositories**.
2. Paste the URL of this repository, set the category to **Integration**, and click **Add**.
3. Find **House Plan** in the list, install it and **restart Home Assistant**.
4. Go to **Settings → Devices & Services → Add integration** and select **House Plan**.
```yaml
resources:
- url:/houseplan_files/houseplan-card.js
type:module
```
The card is registered automatically — no need to add a Lovelace resource manually.
Do not use the on-disk path inside `custom_components`; Home Assistant does not
serve that path as a JavaScript module.
> **Card doesn't load (`Custom element doesn't exist: houseplan-card`) or you manage Lovelace
> resources in YAML?** Add the resource manually pointing at the URL the integration *serves*:
>
> ```yaml
> resources:
> - url: /houseplan_files/houseplan-card.js
> type: module
> ```
>
> Do **not** use `/custom_components/houseplan/frontend/houseplan-card.js` — that is the file
> on disk, which Home Assistant does not serve over HTTP (you'll get a `text/plain` MIME error
> and the element never registers). The correct, integration-served URL is
> `/houseplan_files/houseplan-card.js`. Both cards (`houseplan-card` and
> `houseplan-space-card`) ship in that one file — no separate resource is needed.
### Manual installation
### Manually
Copy `custom_components/houseplan` to `config/custom_components`, restart Home
Assistant, and add the House Plan integration.
1. Copy the `custom_components/houseplan` folder into the `config/custom_components` directory of your Home Assistant.
2. Restart Home Assistant.
3. Add the integration: **Settings → Devices & Services → Add integration → House Plan**.
### Add the card
### Adding a plan screen
Create a new dashboard tab (a "Panel" view works best) and add the card:
Create a dashboard view (Panel works best) and add the card in the UI or as:
```yaml
type:custom:houseplan-card
title:House plan
```
Nothing else needs to be specified — everything else is configured right on the screen.
Different screens may start on different spaces:
---
```yaml
type:custom:houseplan-card
default_floor:ground
```
## How to use
All cards share server-side rooms and coordinates. Current mode, viewport and
selected space remain local to the screen. Revision checks and live sync cover
concurrent clients, but avoid editing the same object in two browsers at once.
### Step 1. Add a space (floor)
## Detailed documentation
On first open the plan is still empty — House Plan immediately offers to create the first space.
If your Home Assistant already has **floors** configured, a wizard offers to create a space
for each floor (names prefilled, a plan image is asked for one by one; any floor can be skipped).

<!-- docs-section: support -->
In the dialog, set a **name** (for example, "1st floor") and pick the background: **upload** a floor-plan image (SVG, PNG, JPG, WebP), **choose one already uploaded** to the server earlier, or select **"no background, I'll draw the rooms"** for a hand-drawn space. The canvas is infinite; an image keeps its own proportions and can be moved and resized at any time in the Background editor.
- Questions and plan examples: [Telegram @ha_houseplan](https://t.me/ha_houseplan).
- Bugs and proposals: [GitHub Issues](https://github.com/Matysh/houseplan-card/issues).
- Before reporting, update House Plan, restart HA and hard-refresh the page.
Include the version, browser, logs and reproduction steps; private entity IDs
may be replaced with fictional ones.
> 💡 You can draw the background in any floor planner (for example, REMPLANNER) or photograph a paper plan. SVG works best — it stays crisp when zoomed in.
Documentation screenshots are produced by the reproducible
`npm run build && node demo/docs/capture.mjs` command using synthetic data only. Scenario version,
source fingerprint and every image hash are recorded in the
[screenshot index](docs/images/screenshots.json).
Later you can add as many spaces as you like (floors, yard, garage) with the **+** button next to the tabs.
### Step 2. Outline the rooms
After the first space is added, the card switches to the **Plan** tab by itself. The card has three mode tabs in the header — **View** (default: display and device control only, nothing can be moved or edited), **Plan** (rooms, openings, labels, space settings) and **Devices** (placing and configuring markers); the edit tabs are shown to administrators. In Plan, click grid points, connecting them with lines, and close the room outline by clicking the first point.
As soon as the outline is closed, the room-save dialog appears. Here you need to **bind the room to a Home Assistant area** — this is exactly what enables the automation. For utility rooms with no devices (hall, sauna) there is a **"No area"** button.

While drawing, a ruler follows the cursor showing the current segment's real length (metres, or feet + inches on an imperial Home Assistant). The scale is set per space — the **"Scale (grid cell size)"** field in the space dialog says how many centimetres one grid cell represents (default 5 cm).
Rooms may not overlap: a click strictly inside an existing room, or an outline that would swallow one, is refused. Two more tools help you reshape the plan later:
- **Merge** — click a room, then a neighbour that shares a wall; they fuse into one. A dialog picks which name and area survive.
- **Split** — click a room, then two points on its walls; the chord cuts it in two. The bigger part stays the room it was (name, area, devices); the smaller one asks for a new name and area.
### Doors, windows and locks
In markup mode the **"Opening"** tool places doors and windows: click next to a wall and the
opening snaps onto it. Pick the type, the **length in real centimetres** (defaults: door 90 cm,
window 120 cm), an open/close sensor and — for doors — a **lock entity**.
With a sensor bound, the plan comes alive: the door leaf swings on its hinge and the swing arc
draws itself in as the real door opens; a window opens its two casements. While open, the moving
parts take an accent colour. A door with a lock shows a padlock badge next to it — green when
locked, orange when unlocked. For safety the lock can **not** be toggled from the plan; a click
on the opening shows a status card with both states instead.
Openings are easy to adjust later: hovering one highlights it, you can **drag it along the
walls** (it slides around corners too), and a **double click opens its properties**.
### Step 3. Devices appear by themselves
As soon as you save a room bound to an area, **the devices of that area are automatically laid out inside the outline**. These are the same devices shown on the **Settings → Devices → (filtered by the room)** page — only the meaningful ones, without service records, bridges and duplicates.
By default only meaningful devices make it onto the plan: non-physical ones (service records, bridges, scenes, individual lamps folded into a light group) arrive with the **"Hide device from plan"** checkbox already ticked. The checkbox is yours from then on — every device dialog has it, virtual devices included. To see and un-hide them, open the device editor and press **"Show hidden"**: hidden devices appear as translucent blue ghosts, a click opens the dialog. Hidden devices still count toward the room's Zigbee signal, but cast no light.
From here on you can just use the plan: clicking an icon opens the device card with the model, link and a button to jump into Home Assistant.

### Step 4. Zoom
The mouse wheel or the **- / ⊹ / +** buttons zoom the plan in and out; on a touch screen the two-finger pinch works. Zoomed out you see the whole plan, zoomed in you see the details, and everything stays crisp. The zoom level is remembered separately for each space.

### Step 5. Put the icons in their places
Switch to the **Devices** tab to arrange icons: drag them with the mouse, click one to open its editor. In **View** mode nothing can be moved — panning the map never displaces a sensor (a top user request). Positions are saved on the server and are identical in all browsers and devices. The **↺** button restores the automatic layout.

### Tap actions: control devices from the plan
By default a tap on an icon opens its info card. In the card settings you can switch
**Tap on a device** to *Toggle* — a tap then switches lights, sockets, fans and
humidifiers directly on the plan (wall-tablet style). For safety, a card-wide toggle
never affects locks, alarms, covers or valves; you can consciously enable toggle for a
specific device (except locks and alarms — those never toggle from the plan) in its
edit dialog. A **long press** always opens the info card.
### Icon rules
Which MDI icon a device gets is decided by **icon rules** — editable right in the card
(the ⬡ button in the header): an ordered list of “name pattern → icon” regexes with a
live test field, bilingual defaults (EN/RU) and a one-click reset. When no rule
matches, the entity *device class* decides (thermometer for temperature sensors, etc.).
### Step 6. Adding your own devices manually
You can also place a **single entity** (not just a whole device): start typing in the binding search and individual entities appear next to devices — handy when one device exposes several values (e.g. temperature and humidity) and you want each as its own icon.
Not everything has to be left to the automation. With the **+** button in the header you can place any device, group or a **virtual point** on the plan (for example, an "Inlet valve" that does not exist as a device). Set a name, icon, model, link, description and, if you wish, attach a **PDF manual**.
The same dialog controls how the device looks on the plan. **Display** switches between the
icon badge, an animated **presence ripple** (pulsing rings while the entity is active, a faint
dot when idle — great for motion sensors) or both, with a per-device ring colour and size. The
**icon size** (×0.5–3) and **rotation** are also per-device, so a wall valve can be small and
turned the way it is mounted.

### Styling the plan with card-mod (advanced, unsupported)
The card ships finished and has no CSS field of its own — but if you already run [card-mod](https://github.com/thomasloven/lovelace-card-mod), every object on the plan now carries a stable hook you can aim at: `data-hp="device"` (plus `data-entity`, `data-area`), `data-hp="room"`, `data-hp="opening"`, `data-hp="decor"`, `data-hp="room-label"`, `data-hp="space-tab"`. We promise not to rename them; we do not ship card-mod, do not support it, and are not responsible for what your CSS does to the card. The full table, the examples and the limits are in **[docs/STYLING-HOOKS.md](docs/STYLING-HOOKS.md)**.
---
## Uninstalling
1. Remove the card (or the tab with the plan) from the dashboard.
2.**Settings → Devices & Services → House Plan → Delete** the integration entry.
3. Remove the integration from **HACS** (or delete the `custom_components/houseplan` folder if installed manually) and restart Home Assistant.
4. Optionally delete the saved plan data: the `config/houseplan/` files (backgrounds and attachments) and the `houseplan.config` / `houseplan.layout` entries in the `config/.storage` directory.
- 📜 [Changelog](docs/CHANGELOG.md) — what changed in every version
([на русском](docs/CHANGELOG.ru.md)).
When reporting a problem, the version number helps a lot: it is shown in the
browser console on load (`HOUSEPLAN-CARD vX.Y.Z`) and in **Settings → Devices &
Services → House Plan**.
---
## Frequently asked questions
**Do I need to write anything in YAML?** No. The only line is adding the card to the dashboard; everything else is done with the mouse.
**My devices did not appear on the plan.** A device appears only if its Home Assistant area is bound to a drawn room. Check that the device has a room assigned (Settings → Devices) and that the room is outlined and bound to that area. If the device exists but is hidden (the "Hide device from plan" checkbox — set automatically for bridges, scenes and other non-physical records) — open the device editor, press **"Show hidden"** and untick the box in its dialog.
**Can I hide an unwanted device or rename it?** Yes — click the device on the plan and press "Edit" in its card: there you can change the name, icon, model or hide the icon.
**Is the data stored in the cloud?** No. Everything is stored locally in your Home Assistant.
---
<p align="center"><sub>Screenshots were taken on a real Home Assistant configuration.</sub></p>
- **Живые состояния и безопасные действия.** Свет и другие безопасные устройства
переключаются с плана; замок нельзя открыть случайным нажатием.
- **Три встроенных редактора.** «План» создаёт комнаты, стены и проёмы;
«Устройства» размещает и настраивает маркеры; «Подложка» добавляет линии,
подписи и мебель.
- **Комнаты, связанные с зонами HA.** Новые устройства появляются автоматически,
а карточки комнат показывают температуру, влажность, свет и средний LQI.
- **Свет и окружение.** Заливки комнат, Glow от ламп, тени от стен, дневной фон и
солнечные лучи из окон.
- **Двери, окна, ворота и пылесосы.** Проёмы отражают реальные датчики и замки;
робот показывает позицию, базу и пройденный путь.
- **Несколько этажей и экранов.** Вкладки пространств, жесты переключения,
локальный масштаб и отдельный стартовый этаж для каждой карточки.
- **Киоск для настенного экрана.** Только план, полноэкранная навигация и размеры
значков, сохранённые отдельно для этого устройства.
House Plan показывает ваш умный дом так, как он выглядит на самом деле — на плане этажей. Вместо длинных списков сущностей вы видите комнаты и устройства на своих местах: где протечка, какая температура в детской, включён ли свет в прихожей, открыты ли ворота.

Это удобно, когда:
<!-- docs-section: first-run -->
- устройств много, и списками пользоваться неудобно;
- нужно быстро понять состояние дома «одним взглядом»;
- хочется отдать доступ близким — по картинке разберётся любой;
- вы хотите красивый обзорный экран для настенного планшета.
## Первая рабочая комната
Интеграция состоит из двух частей, которые ставятся вместе:
1. Установите интеграцию и добавьте карточку на дашборд.
2. Создайте первое **пространство**: загрузите SVG/PNG/JPG/WebP либо выберите
вариант без изображения, чтобы нарисовать план вручную.
3. В редакторе «План» выберите **Контур комнаты**, поставьте вершины и замкните
контур нажатием на первую точку.
4. Назовите комнату и свяжите её с зоной Home Assistant. Для помещения без
устройств выберите «Без зоны».
5. Откройте «Устройства»: устройства связанной зоны уже размещены автоматически;
перетащите маркеры в нужные места.
6. При необходимости оформите подложку линиями, текстом и мебелью.
7. Вернитесь в «Просмотр» — теперь план показывает живые состояния и принимает
безопасные действия.
- **карточка Lovelace** `houseplan-card` — сам интерактивный план;
- **серверный компонент** — хранит разметку комнат и позиции иконок в Home Assistant, поэтому план одинаков во всех браузерах и на всех устройствах.

---

## Чем отличается от аналогов

Обычно план дома в Home Assistant делают через `picture-elements`, `ha-floorplan`
или новые GUI-карточки, где стены и мебель рисуют прямо на дашборде. Там либо
YAML/SVG, либо конфиг живёт в YAML карточки. House Plan — это **общий живой
план** на серверной интеграции Home Assistant:

| | House Plan | picture-elements / ha-floorplan | GUI-рисовалки (напр. easy-floorplan) |
|---|---|---|---|
| **Настройка** | Полностью через интерфейс, мышкой | Ручной YAML / Inkscape SVG | Рисование стен и мебели в карточке |
| **Добавление устройств** | Автоматически по **зоне** HA | Каждую сущность вписываете руками | Ставите сущности руками на чертёж |
| **Координаты иконок** | Перетаскиваете мышью | Считаете пиксели в YAML | Drag на холсте |
| **Разметка комнат** | Встроенный редактор контуров, привязка к зонам | Сторонний SVG-редактор | Сами рисуете стены (мебельный CAD) |
| **Хранение** | На сервере HA (`.storage`, общее, multi-client) | В YAML дашборда | В карточке / YAML дашборда |
| **Оверлеи** | Glow, климат, LQI, солнце, пылесосы, киоск | Что пропишете в SVG/CSS | Зависит от карточки |

**Одной фразой:** House Plan — это общая, area-aware живая карта дома, а не
редактор мебели. Если нужно с нуля нарисовать диваны и стены — лучше
draw-карточка; если есть картинка плана и зоны в HA — House Plan держит один
и тот же layout на всех планшетах.
Пошаговые сценарии, все инструменты и особые случаи описаны в
[полном руководстве](docs/USER-GUIDE.ru.md). Возможности подложки отдельно
зафиксированы в [документе редактора](docs/DECOR-EDITOR.md), а роботов — в
[руководстве по пылесосам](docs/VACUUM.md).
Ключевые преимущества коротко:
- **Никакого кода.** Всё — пространства, комнаты, устройства — настраивается кликами.
- **Автоматическое добавление устройств.** Обвели комнату и привязали её к зоне Home Assistant — устройства этой зоны сами появляются на плане.
- **Ручное добавление своих.** Любое устройство, группу или даже «виртуальную» точку можно поставить на план вручную, задать имя, иконку, модель, ссылку и приложить PDF-инструкцию.
- **Живые состояния.** Температура, уровень сигнала Zigbee, вкл/выкл, открыто/закрыто — всё обновляется в реальном времени.
Цвета значков подчиняются одному принципу — **жёлтый значит «устройство прямо сейчас выполняет свою основную работу»**:
- **Подложку можно двигать и масштабировать.** Картинка плана не прибита к холсту: в редакторе подложки её тянут за тело и за угол, а плашка показывает реальный размер в метрах — чертёж и фотография плана наконец совмещаются.
- **Чёткий зум.** Приближение не «мылит» картинку: план, подписи и иконки остаются векторно-чёткими на любом масштабе.
---
## Настенный планшет / ТВ (киоск-режим)
Отдельный дашборд с view типа «панель», у карточки — `kiosk: true` (или
галочка «Режим настенного устройства» в редакторе карточки):
```yaml
type:custom:houseplan-card
kiosk:true
cycle:0# автосмена пространств каждые N секунд, 0 = выкл (удобно для ТВ)
```
Без шапки и редакторов — только живой план. Свайп листает этажи (при 1:1),
пинч — зум, двойной тап — сброс. Долгое нажатие (3 с) по пустому месту —
настройка размеров значков и текста для ЭТОГО экрана (хранится на
устройстве). Шапку самого Home Assistant скрывают настройки companion-app
или плагин [kiosk-mode](https://github.com/NemesisRE/kiosk-mode).
<!-- docs-section: installation -->
## Установка
В один клик, если у вас уже есть HACS:
### Через HACS
[](https://my.home-assistant.io/redirect/hacs_repository/?owner=Matysh&repository=houseplan-card&category=integration)
[](https://my.home-assistant.io/redirect/hacs_repository/?owner=Matysh&repository=houseplan-card&category=integration)
Если в вашем Home Assistant уже настроены **этажи**, мастер предложит создать
пространство для каждого: названия подставятся сами, план попросит по очереди,
любой этаж можно пропустить.
<!-- docs-section: support -->

## Помощь и обратная связь
В диалоге задайте **название** (например, «1 этаж») и выберите подложку: **загрузите** картинку плана (SVG, PNG, JPG, WebP), **возьмите уже загруженную** на сервер ранее или отметьте **«без подложки, нарисую комнаты сам»**. Холст бесконечный; картинка сохраняет свои пропорции, а подвинуть и отмасштабировать её можно в любой момент в редакторе подложки.
- Вопросы и примеры планов: [Telegram @ha_houseplan](https://t.me/ha_houseplan).
- Баги и предложения: [GitHub Issues](https://github.com/Matysh/houseplan-card/issues).
- Перед отчётом обновите House Plan, перезапустите HA и выполните жёсткое
обновление страницы (`Ctrl+F5`). Приложите версию, браузер, логи и шаги
воспроизведения; приватные entity ID можно заменить вымышленными.

Скриншоты в документации получены воспроизводимой командой
`npm run build && node demo/docs/capture.mjs` только на синтетических данных. Версия сценариев,
fingerprint исходников и хеш каждого изображения находятся в
[индексе снимков](docs/images/screenshots.json).
> 💡 Подложку можно нарисовать в любом планировщике (например, РЕМПЛАННЕР) или сфотографировать бумажный план. Лучше всего SVG — он остаётся чётким при увеличении.
Позже можно добавить сколько угодно пространств (этажи, двор, гараж) кнопкой **+** рядом со вкладками.
### Шаг 2. Обведите комнаты
После добавления первого пространства карточка сама переходит в режим разметки. Кликайте по точкам сетки, соединяя их линиями, и замкните контур комнаты кликом по первой точке.
Как только контур замкнётся, появится окно сохранения комнаты. Здесь нужно **привязать комнату к зоне Home Assistant** — именно это включает автоматику. Для служебных помещений без устройств (холл, сауна) есть кнопка **«Без зоны»**.

Во время рисования у курсора показывается линейка с реальной длиной текущего отрезка (метры или футы+дюймы на имперской системе HA). Масштаб задаётся для каждого пространства — поле **«Масштаб (размер ячейки сетки)»** в диалоге пространства: сколько сантиметров в одной ячейке (по умолчанию 5 см).
Комнаты не могут пересекаться: клик строго внутри существующей комнаты или контур, охватывающий её, отклоняются. Ещё два инструмента помогают перекроить план позже:
- **Объединить** — кликните комнату, затем соседнюю с общей стеной; они сольются в одну. Диалог выбирает, чьё имя и зона останутся.
- **Разделить** — кликните комнату, затем две точки на её стенах; хорда разрежет её надвое. Бо́льшая часть остаётся прежней комнатой (имя, зона, устройства), меньшая просит новое имя и зону.
### Двери, окна и замки
В режиме разметки инструмент **«Проём»** ставит двери и окна: кликните рядом со стеной — проём
примагнитится к ней. Выберите тип, **длину в реальных сантиметрах** (по умолчанию дверь 90 см,
окно 120 см), датчик открытия и — для двери — **замок**.
С привязанным датчиком план оживает: створка двери поворачивается на петле, и дуга распахивания
дорисовывается по мере открытия настоящей двери; окно раскрывает две створки. Пока открыто,
подвижные части подсвечены акцентным цветом. У двери с замком рядом отображается замочек —
зелёный, когда заперто, оранжевый, когда нет. Ради безопасности замок с плана **нельзя**
переключить — клик по проёму показывает карточку с обоими статусами.
Проёмы легко поправить позже: при наведении проём подсвечивается, его можно **перетащить вдоль
стен** (в том числе за угол), а **двойной клик открывает свойства**.
### Шаг 3. Устройства появляются сами
Как только вы сохранили комнату с привязкой к зоне, **устройства этой зоны автоматически расставляются внутри контура**. Берутся те же устройства, что показаны на странице **Настройки → Устройства → (фильтр по нужной комнате)** — только осмысленные, без служебных записей, мостов и дубликатов.
По умолчанию на план попадают только осмысленные устройства: нефизические (служебные записи, мосты, сцены, лампы, свёрнутые в световую группу) могут быть скрыты автоматически. Управление находится в левом нижнем углу диалога устройства: **«Скрыть»** убирает маркер после сохранения, а у уже скрытого маркера там же появляется **«Показать»**. Чтобы найти скрытые устройства, откройте редактор устройств и нажмите **«Показать скрытые»**: они отобразятся полупрозрачными синими призраками, клик открывает диалог. Скрытые устройства учитываются в Zigbee-сигнале комнаты, но света не дают.
Дальше можно просто пользоваться планом: клик по иконке открывает карточку устройства с моделью, ссылкой и кнопкой перехода в Home Assistant.

### Шаг 4. Масштаб
Колесо мыши или кнопки **- / ⊹ / +** приближают и отдаляют план; на сенсорном экране работает «щипок» двумя пальцами. При отдалении виден весь план целиком, при приближении — детали, и всё остаётся чётким. Масштаб запоминается отдельно для каждого пространства.

### Шаг 5. Расставьте значки по местам
Расставлять значки нужно на вкладке **«Устройства»**: там они перетаскиваются мышью, а клик открывает редактор. В режиме **«Просмотр»** ничего сдвинуть нельзя — панорамирование карты больше не сдвигает датчики (главная просьба пользователей). Позиции сохраняются на сервере и одинаковы во всех браузерах и устройствах. Кнопка **↺** возвращает автоматическую раскладку.
прямо в карточке (кнопка ⬡ в шапке): упорядоченный список «шаблон имени → иконка»
с живым тест-полем, двуязычные умолчания (EN/RU) и сброс одной кнопкой. Если ни одно
правило не подошло — решает *device class* сущности (термометр для датчиков
температуры и т.п.).
### Шаг 6. Добавление своих устройств вручную
Можно поставить и **отдельную сущность** (не только устройство целиком): начните печатать в поиске привязки — рядом с устройствами появятся отдельные сущности. Удобно, когда одно устройство отдаёт несколько значений (например, температуру и влажность), а вы хотите каждое своей иконкой.
Не всё нужно оставлять на автоматику. Кнопкой **+** в шапке можно поставить на план любое устройство, группу или **виртуальную точку** (например, «Вентиль на вводе», которого нет как устройства). Задайте имя, иконку, модель, ссылку, описание и при желании приложите **PDF-инструкцию**.
В этом же диалоге настраивается вид устройства на плане. **Отображение** переключает значок,
анимированную **пульсацию присутствия** (расходящиеся кольца, пока сущность активна, и тусклая
точка в покое — идеально для датчиков движения) или то и другое сразу, с цветом и размером колец
на устройство. **Размер значка** (×0,5–3) и **поворот** — тоже индивидуальные: вентиль на стене
может быть маленьким и повёрнутым так, как он установлен.

### Свои стили через card-mod (для продвинутых, без поддержки)
Карточка приезжает готовой, и поля для CSS у неё нет — но если у вас уже стоит [card-mod](https://github.com/thomasloven/lovelace-card-mod), у каждого объекта плана теперь есть стабильный «крючок», за который можно зацепиться: `data-hp="device"` (плюс `data-entity`, `data-area`), `data-hp="room"`, `data-hp="opening"`, `data-hp="decor"`, `data-hp="room-label"`, `data-hp="space-tab"`. Мы обещаем их не переименовывать; сам card-mod мы не поставляем, не поддерживаем и за то, что ваш CSS сделает с карточкой, не отвечаем. Полная таблица, примеры и ограничения — в **[docs/STYLING-HOOKS.md](docs/STYLING-HOOKS.md)**.
---
## Удаление
1. Уберите карточку (или вкладку с планом) из дашборда.
2.**Настройки → Устройства и службы → House Plan → Удалить** запись интеграции.
3. Удалите интеграцию из **HACS** (или папку `custom_components/houseplan` при ручной установке) и перезапустите Home Assistant.
4. При желании удалите сохранённые данные плана: файлы `config/houseplan/` (подложки и вложения) и записи `houseplan.config` / `houseplan.layout` в каталоге `config/.storage`.
---
## Помощь и обмен опытом
- 💬 **[Чат в Telegram — @ha_houseplan](https://t.me/ha_houseplan)** — вопросы,
помощь с настройкой, идеи и скриншоты ваших планов. Самый быстрый способ
связаться с автором и другими пользователями.
- 🐞 [Issues на GitHub](https://github.com/Matysh/houseplan-card/issues) — баги
и запросы фич (пожалуйста, указывайте версию House Plan).
- 💡 [Discussions](https://github.com/Matysh/houseplan-card/discussions) — для
развёрнутых обсуждений.
- 📜 [История изменений](docs/CHANGELOG.ru.md) — что менялось в каждой версии.
Версия видна в консоли браузера при загрузке (`HOUSEPLAN-CARD vX.Y.Z`) и в
**Настройки → Устройства и службы → House Plan** — с ней разбираться сильно
быстрее.
---
## Часто задаваемые вопросы
**Нужно ли что-то писать в YAML?** Нет. Единственная строчка — это добавление карточки на дашборд; всё остальное делается мышкой.
**Мои устройства не появились на плане.** Устройство появляется, только если его зона в Home Assistant привязана к нарисованной комнате. Проверьте, что у устройства задана комната (Настройки → Устройства), а комната обведена и привязана к этой зоне. Если устройство есть, но скрыто автоматически — откройте редактор устройств, нажмите **«Показать скрытые»**, откройте его диалог, нажмите **«Показать»**, затем **«Сохранить»**.
**Можно ли скрыть лишнее устройство или переименовать его?** Да — кликните по устройству на плане и в его карточке нажмите «Редактировать»: там можно сменить имя, иконку, модель или скрыть значок.
**Данные хранятся в облаке?** Нет. Всё хранится локально в вашем Home Assistant.
---
<p align="center"><sub>Скриншоты сделаны на реальной конфигурации Home Assistant.</sub></p>
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.