The 2 ms reset of ESPHome's expander support (0.2.81) left the chip silent: "Calibration error" with the PCA9557
answering at 0x18. A component of smart_display's (elecrow_touch_reset.h, setup priority 700: after the I2C bus,
before the GT911 driver) writes the expander as Elecrow's V3.0 ESPHome example does: every pin an output and low,
20 ms, the reset released, 100 ms, the interrupt line back to an input. The driver's reset and interrupt pins are gone.
A unit's touch never answered: "touchscreen is marked FAILED: Calibration error" at setup, the GT911 still in reset
behind the PCA9557 (I2C 0x18) of the V3.0 board. The board file declares the expander as ESPHome's pca9554 (the same
register map) and hands the GT911 driver its reset (IO0) and interrupt (IO1) pins on it, so the driver runs Elecrow's
own sequence: reset low with the interrupt line low, reset high, 56 ms, address 0x5D. A V2.0 without the expander boots
its GT911 by the pull-up as before. The bus is scanned at boot for the log.
The GT911 of the CrowPanel 7.0" sits on GPIO 19 and 20, the ESP32-S3's USB pins. Firmware 0.2.79 freed them once, in a
static constructor; the Wi-Fi PHY library then kept the USB PHY on (CONFIG_ESP_PHY_ENABLE_USB, on by default on an
ESP32-S3) and the touch went quiet once Wi-Fi came up. The board file turns that option off, frees the pins again in
BOOT_TOUCH after Wi-Fi started, and a one-second watchdog frees them whenever anything takes them, with a log line.
A board file for the Elecrow CrowPanel 7.0" (DIS08070H, ESP32-S3-WROOM-1-N4R8): 800x480 through mipi_rgb, GT911 on the
ESP32-S3's USB pins (elecrow_usb_phy.h frees them, behind a flag only this board sets), the Guition's vertical sizes
across a wider page, no rotation. Its 4 MB of flash take the CYD's budget: a partition table of the component's
(smart_display: partitions: 4mb_wide_slots, 1,984 KB slots), the icon fonts at 2 bits a pixel and the TLS client only
bring the firmware to 1,865,440 bytes, 91.8 % of a slot, with camera images kept. docs/ELECROW.md has the numbers and
the trade.
The add-on reads a large board's name from the type sensor's value, the editor offers the board under New screen and
draws the wide preview, tools/check.sh builds it with the budget, render_topbar and capture_ui take the panel's size.
Firmware 0.2.79; the CYD image is unchanged (1,633,760 bytes).
This fork builds from gittea.trippedonmy.name/Xandrossian/homeassistant_espscreen: the packages, the fonts, the
component and the add-on repository point there instead of GitHub.
A single or double-width media player tile takes Display -> Album cover
on a Guition: the cover of what plays in the icon's place, like Home
Assistant's own tile. It rides in the page's strip of 0.2.91: the request
and the answer are the same, a media player is simply among the tiles.
The app fetches a player's picture only when its address changes; a page
with a camera loads at the camera's pace with the cover reused, a page of
media tiles alone loads once, and a new track (the picture's mark in the
state, part of the page's wish) asks again at once. A player without a
picture keeps its icon and is not asked for; the double-width tile keeps
its controls; the tile over the whole page keeps the card's big cover.
Firmware 0.2.78.
Host harness 31 checks (docs/TEST_RESULTS_0292.md). CYD firmware
1,633,760 B, 89.0 %.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Live picture (issue #4): a camera or image tile on a Guition takes
Display -> Live picture with a pace of 15 or 30 s. The camera tiles of a
page share one download: the screen asks ESP Screens for them together
(esphome.screen_camera with tiles, size and bg) and the app serves one
BMP, a strip of squares with the corners rounded over each tile's own
colour; every tile draws its square out of it with LVGL's image offset
(LV_IMAGE_ALIGN_TOP_LEFT: the default centres a taller source). A third
online_image on the Guition holds the strip; it loads after the alert's
picture, a cover and the camera full screen, never under a card, in
standby or under a finger. The strip is served whole every time: ESPHome's
http_request logs a 304 as a failed request. Firmware 0.2.77.
Alert button: esp_screens_show_alert takes action and data; the app
performs the action once when the button's report (esphome.screen_alert,
ok) comes from any screen. show_alert keeps its seven fields, so every
firmware from 0.2.31 has it.
Media card and full-page media tile: a title or artist line that is too
wide rolls by (LV_LABEL_LONG_SCROLL_CIRCULAR) instead of ending in dots.
Full-page tiles are white or their own pastel like every other tile; the
tint in the state colour of firmware 0.2.62 to 0.2.76 is gone.
Host harness (27 checks) and bench Guition + local add-on on the Yellow:
docs/TEST_RESULTS_0291.md. CYD firmware 1,633,728 B, 89.0 %.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The add-on ships ESPHome 2026.9.0: it builds an ESP32 with ESP-IDF itself
and keeps a compiler cache. firmware.py gives the CLI its own storage in
/data (ESPHOME_DATA_DIR /data/esphome, which the ESPHome Device Builder no
longer wipes on start; ESP-IDF and ccache in /data/idf, at most 1 GB;
compilers limited to the core count as PlatformIO ran them), removes the
unused PlatformIO once before the first ESP-IDF build, and leaves the new
folders out of backups.
Firmware 0.2.75: assertion_level SILENT in every board's esp32 block, the
core logs from INFO, and clock_parts() in runtime_model.h replaces sscanf.
timer_left() caps a running timer at its duration (the screen's clock runs
up to a second behind Home Assistant's). The packages keep min_version
2026.6.2, the top-level online_image and OTA passwords for new screens, so
an add-on or Device Builder that isn't updated yet keeps building every
screen; CI now builds both boards with the add-on's ESPHome and with
min_version. ESP-IDF's error names stay on, and build_flags stay under
platformio_options, because ESPHome 2026.9's LVGL reads only those.
tools/check.sh: 478 Python tests, 19/19 C++, editor; --firmware PASS with
ESPHome 2026.6.2 (CYD 87.7 %), 2026.8.1 (88.3 %) and 2026.9.0 (86.4 %,
was 91.5 %). Both bench screens OTA'd and watched for half an hour;
ESP Screens as a local add-on on the Yellow built both profiles
(CYD 1,037 s first build, 41 s unchanged). Details:
docs/TEST_RESULTS_0289.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
packages/core.yaml sets wifi: power_save_mode: none. ESPHome merges it
under the screen's own wifi: block, which keeps the network, the password
and the hotspot, and the screen's own YAML still wins on the mode. The
YAML ESP Screens writes has had the line since app 0.2.24; screens with
an older or hand-written YAML ran ESPHome's ESP32 default light, and
their messages from Home Assistant waited 76-117 ms on the bench.
esphome config on the bench screens' Device Builder YAMLs: NONE for both
without a line of their own, LIGHT when the YAML says light. The bench
Guition, flashed from its Device Builder YAML: ping 15 ms average instead
of 117 ms. tools/check.sh: 474 Python tests (a new one keeps the core's
wifi: block to this one key), 19/19 C++, editor; firmware unchanged in
size (CYD 1,678,816 B, 91.5 %; Guition 2,055,280 B), since the checkout
profiles already had the line.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Starting screen: until the first layout, render() keeps the status text
("Connecting to Home Assistant", "Waiting for ESP Screens") and a spinner
as one block in the middle of the page; the first layout deletes both.
The room label stays empty meanwhile, and ROOM_NAME defaults to "".
Camera: the full screen shows a spinner instead of "Loading image". The
screen asks when the camera opens and loads a link as soon as it comes
(camera_load, never under a finger) instead of waiting for the 250 ms
tick twice, and online_image reads 16 KB per loop instead of 4 KB. On the
bench Guition with the EZVIZ camera: cold 5.8 -> 4.55 s, warm 3.4 -> 2.06 s.
Alert picture: the image takes the frame's radius (18, the alert card's);
LVGL's software renderer clips it row by row, without a layer.
One spinner_create() makes the busy sheet's, the starting screen's and the
camera's spinner; theme role CAMERA_TRACK colours the camera's ring. The
README gets a row of camera pictures; docs/CAMERA.md, RELEASING.md and
TEST_RESULTS_0287.md describe the measurements.
Checked with tools/check.sh (473 Python tests, 19/19 C++, editor) and
--firmware: CYD 1,678,816 B (91.5 %, +1,312 B), Guition 2,055,280 B
(25.3 %). Host renders of both boards in both looks; four OTA builds on
the bench Guition Wallbox, which runs this firmware.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A tap on the On/off switch of a wide or full-page tile moved the knob to on
and back, and the light stayed off. Since firmware 0.2.59 control_event drew
the new position first (Tile::optimistic writes it into state) and then asked
key_action for the action, which read that new position: an off light got
light.turn_off, an on light light.turn_on. A tap on the tile itself was never
affected.
tile_controls::press_key chooses the action first and then shows the new
position; control_event uses it. Unit tests cover a light, a fan, a sensor
and a cover key.
Checked with tools/check.sh (471 Python tests, 19/19 C++) and compiled for
both boards: CYD 1,677,504 B (91.4 %, +16 B), Guition 2,054,640 B (25.3 %).
On Max's Guition Wallbox (OTA from this branch) every tap on the switch sent
the right action and Home Assistant's logbook followed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The airco tile stayed orange while the airco was off. The firmware kept its own
short list of what turns grey and gave an off climate Home Assistant's orange,
but Home Assistant has no colour for a climate that is off: its cards fall back
to the inactive grey. Other tiles had the same fault: an off fan, a closed
blind, a docked robot, a player in standby, an idle script, a paused timer and
a camera that isn't streaming all kept their colour.
Tile::active() is now a port of the frontend's stateActive() for the domains a
tile can show, and every tile that is not active is grey. An airco that is off
reads Off, with the room's temperature when the device reports one, as Home
Assistant's tile does. tile_controls::accent() follows stateColorCss(): alarm
binary sensors red, a battery green, orange or red by its charge, a running
script or timer and a streaming camera amber, the weather per condition, the
sun amber and indigo, someone in a zone blue. Scenes, selects, numbers and
sensors keep their own colour per kind. A full-page tile lights up only for
tiles that can be off (Tile::lights_up), and a closed blind's slider keeps its
colour. The top bar's alarm classes are Home Assistant's eleven.
Checked with tools/check.sh (471 Python tests, 19/19 C++ with 58 state cases
and every colour rule) and compiled for both boards: CYD 1,677,488 B (91.4 %,
+576 B), Guition 2,054,640 B (25.3 %).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The two board profiles (5,000 lines each, 87 % the same text) become packages/core.yaml and one file per
board under packages/boards/: the hardware, a table of 84 sizes and font sizes, 18 hooks with the lines of C++
only that board adds inside a shared lambda, and the board's own sections. packages/cyd.yaml and
packages/guition.yaml keep their names and places (a screen installed from ESP Screens builds as before);
home-like-2432s028.yaml and guition-4848s040.yaml become the same entries for a checkout.
The firmware stays 0.2.70: the C++ ESPHome generates for each board, through the checkout profile and
through the GitHub package route, was compared line by line with the old profiles' (docs/TEST_RESULTS_0284.md).
tools/check_packages.py replaces tools/generate_packages.py; tools/profiles.py gives the tests and tools the
three files as one; docs/PROFILES.md says how the files fit together and how a board is added.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A light that offers effects gets a sparkles key at the top right of its colour
card. It opens a page with a row per thing the lamp offers, all from Home
Assistant: the light's effect and the select entities of its device (a WLED's
colour palette, preset and playlist) with what they are set to, and a slider per
number entity of the device (speed, intensity). Names are Home Assistant's own
English entity translations, icons its frontend's; nothing about a brand is
hardcoded beyond a preferred order of translation keys.
A row opens LVGL's roller with every name Home Assistant lists at that moment,
asked for when the picker opens (esphome.screen_options, answered with
op: options, one page of names per message), the effects alphabetically with
Solid on top. The check key at the top right sends the choice as one action;
the back key sends nothing. A slider sends number.set_value when let go. While
an effect runs, the tile names it instead of the brightness.
Firmware 0.2.70: effects_page.h (built when it opens, gone when it closes),
Extra.effect/option_rows/number_rows, options_request, a hidden roller widget
so ESPHome compiles LVGL's roller, five firmware-only glyphs. App 0.2.83:
light_effects.py, the effect attribute, the device rows in the state message,
card_options_loop. Checked on the host with a virtual finger (35/35 on both
boards) and compiled for both ESP32 boards; CYD 1,675,424 B (91.3 %).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
GitHub issue #9 asked to drop the Previous/Next bar when Go to page tiles do the navigation and to give the tiles the whole screen. Page buttons is a setting the screen owns (on by default): a row in the Screen group of the settings page, switch.<screen>_page_buttons in Home Assistant, and a row in the editor's Screen settings. Without the bar, and on a screen with one page, place_page moves and sizes the three rows so they reach the bottom margin (Guition 108 -> 122 px, CYD 52 -> 61 px); a card keeps its circle, name and state in its middle and the single analog clock keeps its dial. With page buttons and swiping both off, the editor says where the Go to page tiles lead and which pages that leaves out of reach.
The bar itself: each half of the band under the tiles is one key with an MDI chevron and lights up under a finger, and a dot per page sits between them; the screen's settings page uses the same pager. App 0.2.82, firmware 0.2.69.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
On a CYD the strip at the bottom of a compact card is 8 px high, so a finger landing a little above it pressed the card: the light's card opened, the light toggled, or the page-swipe gesture turned the page without a log line. The strip's click area now covers its card (slider_zone); a press that never moves forwards the card's tap or hold and shows the card pressed; the CYD's page-swipe handler and the Guition's flick handler leave a captured slider alone and log what they do; a slider commit no longer bounces against its own previous send (TouchGuard::accept_within); the CYD's edge band is 20 px because its resistive panel reports up to the bezel. App 0.2.81, firmware 0.2.68.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A fast swipe off the right side of a Guition ended at 93-98 %: the GT911 reports the finger for the last time 33-48 px inside the edge while a wide tile's slider ends 29 px from it, and LVGL keeps the last point it saw. cyd::edge_snap() takes a release inside the band (SLIDER_EDGE_SNAP_PX: Guition 67, CYD 56) as the end that lies in that band, for the tile sliders and the light card's sliders; the log names each snap. App 0.2.80, firmware 0.2.67.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
On the tile (and the strip of a full-page blind) the small slider filled with the open part, while the blind's card
draws the blind hanging from the top: a closed blind looked open on one and closed on the other. Now both fill with
the closed part and dragging right closes; the text still gives the position (open %). The editor's mockup follows.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A test and code-analysis round of 0.2.72 to 0.2.76 (suites at every tag, sanitizers, ESP32 and GitHub builds, host
harnesses on both boards, the editor in a browser, best practice checked against ESPHome, LVGL, Home Assistant and
Vue) found these, fixed here with Max's choices for the product questions:
- Firmware 0.2.65: the second hand of a single analog clock moves again; on a page that is one full on/off tile the
waking push also switches it; on the Guition a flick is never a tap and a finger that slides off a tile doesn't tap
it; Previous/Next react to a firm press; the same Go to page tile may sit on several pages; a layout the screen has
no memory for is refused instead of crashing; 64-bit redraw mask for 48 tiles; Montserrat 14 no longer built.
- App: Save refuses a layout too long for one message or too big for eight pages; an offline screen keeps its
firmware version from the device registry; HA refusals reach the editor; 128 KB requests; changelog out of the
live stream; cached hashed assets; tini as init and a clean stop; changelog parsing without asterisks or crashes.
- Editor: clicking the open screen keeps unsaved edits, edits during a save stay unsaved, a Page row and phone
dragging to other pages, repeated page tiles, limits from the server, HA's error messages, a type-checked build.
- tools/check.sh + GitHub Actions + a CYD flash budget; docs match 48 tiles and the new editor.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A download shares ESPHome's loop with touch and drawing. An alert already
closes every card, and the card's cover with it; a media tile over the whole
page stays under the alert. Its cover now waits while the alert's picture is
announced, loading or due for a retry, and an alert picture that arrives while
a cover is on its way (or a dropped cover's download is still being closed)
starts right after it. The host harness checks all three cases on the Guition.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The card behind a media tile shows the album cover, the title, the artist and
the album, a progress bar that runs while the track plays with the times at its
ends, round keys for previous, play or pause and next, and a volume row with a
mute key, a wide slider and the percentage. Keys the player lacks stay faded,
an off player shows one power key, a stream without a length shows no bar, and
the CYD shows the player's icon in the cover's place. A media tile over the
whole page is the same card under the tile's head. media_card.h holds the
geometry, tests/test_media_card.cpp checks every shape.
The cover comes the way camera images do: the screen asks with the size and
the colour behind the corners, the app fetches the picture from Home Assistant,
cuts it square, rounds the corners over that colour and serves it on port 8098;
the Guition loads it once and asks again only when the picture's mark in the
state changes. Firmware 0.2.64.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Both boards draw a normal tile with the icon on the left and the name and value beside it;
the Vue editor stacked icon, name and value in one column. On a CYD mockup a tile is 47 px
high and that stack needed 58 px (63 for a watch card), so the value line was cut in half and
a big value hid the name. A single tile is now a row (.head: icon + .tx), a watch card puts
the name row on top and the number under it, and the small slider runs underneath; on a CYD
a tile with a slider uses the screen's 4 px padding and 8 px slider. The Guition mockup
follows the screen the same way.
App 0.2.76, firmware stays 0.2.63. Test results in docs/TEST_RESULTS_0276.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The Material Design Icons fonts of both boards had no bpp, so ESPHome rendered them at
its default of one bit per pixel while the text fonts already had four. On the CYD, at
18 and 28 px, a thin stroke such as the ring of mdi:power came out lumpy and uneven, and
every other icon had a ragged edge. All four icon fonts of both profiles now say bpp: 4;
the icon generator and its test keep that line. Costs the CYD 95 KB of flash (90 % of
the update slot in use on top of 0.2.74), no RAM.
App 0.2.75, firmware 0.2.63. Test results in docs/TEST_RESULTS_0275.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Max asked for many more tiles, a tile that takes the whole page as a light
switch you push without looking, and tiles that go to a page. The limit is
one tile per slot of the eight pages, 48, and the firmware keeps its tile
list on the heap sized to the layout (PSRAM on a Guition). A tile's size can
be `full`: one big button that lights up in the state colour while on, or the
double-width card's head with its small slider, direct controls or graph at
the bottom and the value the double-width card had no room for in between;
the clock, the forecast (with an hours strip) and the sun path fill the page.
Navigation tiles are the built-in entities screen.page_1 to screen.page_8:
an arrow or a chosen icon, the name and "Page n", a tap opens that page.
The app sends the three only to firmware 0.2.62+, the editor (the Vue port of
this round by the editor session, 00c9a4e) offers Normal / Double-width /
Full page and "Goes to page", and Claude in Home Assistant gets size: full and
the page tiles. Built-in tiles now send a chosen icon as its codepoint.
Found and fixed by the new host harness before release: the one-second tick
moved the analog clock's second hand (part 18) on a full forecast's hour label
during a page switch and corrupted it. Firmware 0.2.62; both C++ and Python
suites, host harness 37/37 on both boards, hardware pass on both bench boards
in docs/TEST_RESULTS_0274.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The tiles draw what Home Assistant reports (a new /api/states answers for the tiles on the page); the ···
menu gets Identify (the screen's own show_alert with flash), Copy layout from another screen, Export and
Import as JSON; the Alerts page opens with Try it (/api/alerts/test, one screen or all); the library filters
by room and hides what is placed, with lit icons for what is on; ⌘K searches screens, entities and actions;
the Update badge lists what's new from the changelog (copied into the image) and a running update shows a
progress bar with the ESPHome stage and the last log line.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The page is now web/ (Vue 3 + Vite + TypeScript), built into screen_manager/app/static
with hashed file names; server.py serves index.html as built and /assets/. The Python
API, the data and the firmware are unchanged. Your screens sit in a sidebar, the pages
of a screen stand next to each other at the screen's own shape, the library sits on the
right, and a tile's settings slide in as a drawer while the mockup stays in view. New
screen, Firmware & USB, Alerts, Override YAML and Settings are pages of their own.
Fonts moved to web/src/assets; the tests read the source through tests/editor_sources.py.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The Previous and Next bars ran their tap through the tile guard, whose 600 ms
same-tile window (so one contact never fires a tile twice) dropped the second
and third tap of a quick Next, Next, Next. They now use accept_repeat, the
-/+ keys' rule since 0.2.25: every clean tap counts, only a contact within
150 ms of the last accepted tap on the same button is a bounce. The page fill
already dropped a fill under way when the next page was asked for. Firmware
0.2.61; test_cyd_ui.cpp covers the page buttons; host harness and check builds
in docs/TEST_RESULTS_0272.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A light group's lamps fade, Home Assistant reports every step, and the tile's slider followed each one: after a drag it
jumped back to the old value and crept up. The Tile now holds the sent value (brightness, fan percent, volume) until
Home Assistant reports within 3 % of it, reports the entity off, refuses, or reports once and stays quiet for 1.5 s;
never longer than 8 s, and without any report the hold ends with the wait. A cover's position slider keeps following
the blind. Firmware 0.2.60; host checks 17/17 on both boards.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Home Assistant answered every command in 342-599 ms on a real installation,
yet a tile was busy for at least a second and up to six without a confirming
state. Now nothing is drawn for the first 400 ms, a tap that switches shows
the new stand at once as Home Assistant's own switch does, and the sheet
appears only for a slow command. Home Assistant's "it worked" answer ends the
wait 800 ms later, which covers a stop on a cover that already stands still,
and a wait without any answer ends after three seconds instead of six. The
cards' keys, sliders, chips and the -/+ edit now ask for an answer too, so a
refusal shows wherever a finger presses.
Firmware 0.2.59. A 200 ms LVGL timer runs only while a command is under way.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
rounded_state raised decimal.InvalidOperation for a value beyond Decimal's
28-digit context, such as 1e30; the exception escaped state_message and the
whole screen's sync stalled. Such a value now passes through unrounded.
App only, firmware stays 0.2.58. Found in a review of 0.2.67.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
GitHub issue #8: Home Assistant on a server or virtual machine is out of
reach of the screen. New screen and Firmware & USB get a Download choice:
ESP Screens builds the profile's firmware as always and offers ESPHome's
factory image, which the owner flashes with ESPHome Web in Chrome or Edge.
- Firmware: a download job compiles without uploading; only the image of
the profile's latest successful build in this process is served, a new
job withdraws it, and a build without a factory image fails.
- USB on the Home Assistant machine stays first in both lists, also before
a board is plugged in; Download and Later stay chosen once picked.
- The not-yet-in-Home-Assistant card knows a downloaded screen.
- The device name pattern is valid again under Chrome's v flag.
- Docs: README, DOCS, EASY_SETUP, DOCKER, TROUBLESHOOTING. App only;
firmware stays 0.2.58.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
GitHub issue #7 asked for cover.toggle on tap. ESP Screens now asks Home
Assistant which actions each entity has instead of keeping lists: the editor
offers On / off (covers included), sliders, controls and displays only where
Home Assistant has the action, and refuses only new unsupported settings.
Perform action runs any action Home Assistant offers for the tile's entity,
under its own names and fields; the screen sends it with a call id and says
Refused when Home Assistant refuses. Sensors round by display precision, and
tiles, the top bar, timelines, vacuum chips and the inspector use Home
Assistant's words where they showed raw states. Tiles show Home Assistant's
icons, picked as its frontend does (the fonts grow to 309 glyphs).
Existing screens keep every setting and tap as before: stored layouts load
leniently, legacy taps route the same (1020 combinations), and messages are
byte-identical without Home Assistant's words and icons. Firmware 0.2.58.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
An alert from esp_screens_show_alert takes a `camera` (a camera.* or image.* entity): a Guition with firmware
0.2.57 shows that moment's picture across the top of the card, and a tap opens the camera full screen. A camera
or image tile opens the same full-screen view, refreshed every four seconds at a steady pace.
ESP Screen Manager fetches the snapshot with its own access, sizes it for the screen and serves a 24-bit BMP
under a random, short-lived link on port 8098; the screen loads it with ESPHome's online_image and holds no
Home Assistant token. BMP because ESPHome decodes it per chunk, where a full-screen JPEG held the loop 0.6 s.
The CYD keeps its alerts without the picture and gets no camera tiles.
The Guition also builds with ESPHome 2026.8+ again: backlight_fade.h derived from ESPHome's LEDC output, which
2026.8 made final, so the backlight now dims through ESPHome's own light transition.
Tested on the bench boards and a local add-on on the Home Assistant Yellow; docs/TEST_RESULTS_0266.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Wake from Home Assistant counted as a tap: it reset last_touch_ms, the one
clock that standby and Back to page 1 both read. An automation pressing Wake
on every motion therefore held an open card or page 2 up for as long as
someone moved in the room. Back to page 1 now has its own clock, last_use_ms,
written only by the boot, the touchscreen triggers, open_settings from Home
Assistant and the UI self test. Wake, an alert ending and Auto standby on
still restart the standby time, and Wake still brings back the brightness and
the second hand.
The rule moved from the interval into the script back_to_page_1_when_due.
Wake runs it right after waking a dimmed screen, so a page whose time ran out
during standby lights up on page 1 instead of jumping there a second later.
ESP Screens no longer opens the first screen by itself. Until one is picked
the page says "Choose a screen", and with no screens it still offers the
install; both cards start hidden so neither flashes while the page loads.
Checked with 48 host API checks per board, both ESP32 builds, 314 Python and
16 C++ tests, and the editor on the demo home. Firmware 0.2.56.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The title label had long_mode: DOT and a width but no height. LVGL 9.5 only
puts the dots on a label whose wrapped text is taller than the label, so the
label grew with its lines and a long title ran into the subtitle. It now has
one line of the headline font as its height (32 px Guition, 21 px CYD).
The Claude skill and the editor's alert tips said the CYD shows about forty
characters of a title; both screens show about twenty.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Dark mode is a new switch under Brightness on the screen's settings page, in
the Screen settings of ESP Screens and as switch.<screen>_dark_mode in Home
Assistant, so an automation can turn it on at bedtime. The screen changes
look at once, open cards and an alert included, and keeps it after a
restart. The design stays the same: a black page, graphite cards with a
hairline, soft white text, Home Assistant's state colours where they tell
something, and pastel card colours as deeper versions of themselves.
Every colour the firmware chooses now lives in components/smart_display/
theme.h: roles with a light and a dark value, the named card colours and
Home Assistant's state colours. The board profiles give widgets colourless
shared styles that theme.h fills, so a change of look refills those styles
and draws again what code painted by state. The light look is unchanged
apart from merges below one panel step. Dark mode costs 440 bytes of static
RAM on the CYD, and the LVGL tree of the host build is 416 bytes smaller; tests
keep colours out of every other file.
Comparing a screen that switched looks with one that booted in it also
found two older bugs, now fixed: a settings switch that was on showed its
knob at the left whenever its page was drawn, and the sun card's sunlit
area turned orange after a palette change. Firmware 0.2.54.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A binary sensor's tile and card now say what Home Assistant says for its
device class, such as Open or Closed for a door and Wet or Dry for a leak
sensor. The screen maps the class it already receives, with the same words
as the top bar and the history card; a test keeps both tables equal. A
history card still waiting for its answer no longer shows the previous
card's word.
Lights and binary sensors that are off turn grey, as in Home Assistant, and
a light without an icon of its own shows a crossed-out bulb while it is off,
on the tile and in the top bar. A chosen icon stays and turns grey.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Override YAML on a screen edits <name>.local.yaml, a package loaded after
the shared board package. New profiles get an empty one; an existing
profile is attached on the first save without reformatting the file. The
editor validates YAML, refuses the managed sections and substitutions,
and Save & check runs ESPHome's full validation of the complete profile.
The example uses !extend: ESPHome appends package lists instead of
merging them by id, so a bare id added a second display without a
platform and failed the build. An override is 12 KB at most and a
too-large body answers as JSON instead of a bare 413. The sidecar is
written before the profile includes it, and a symlinked sidecar is
refused. Firmware stays 0.2.52.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Since firmware 0.2.42 a slider showed the tile's colour only while
Tile::active() held, which is never true for a playing, paused or idle
media player, a cover or a number, so their volume, position and value
sliders stayed grey like a light that is off.
Tile::slider_active() follows Home Assistant's tile sliders instead: grey
for an unavailable entity, a light or fan that is off and a media player
that is off or in standby; a closed cover keeps its colour as in HA's
cover position feature, and a number with a value is active. The sliders
of direct controls on wide tiles and the small sliders use it;
Tile::active() and its other uses are unchanged.
Checked with 282 Python and 15 C++ tests, generate_packages --check, ESP32
builds of both check profiles and host renders of both boards.
Firmware 0.2.52.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A sensor, number, binary sensor, person or held switch opens a history card:
a line with round axis steps, clock times and the highest and lowest moment
for numbers, a timeline with the time in each state for on/off, zones and
statuses. The screen asks with esphome.screen_history for 1 hour, 24 hours or
a week; the app answers with one compact history message from the recorder's
statistics or the state changes. A finger on the graph reads the value and
time of that moment at the top while the graph stays as it is. The Guition
weather card starts below its round back button, and the README shows the new
cards and the light and weather cards as they look now.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A tap on a cover now opens a card like Home Assistant's own, in the style
of the climate and vacuum cards: a tall position slider on which the blind
hangs from the top, a tilt slider over slats for venetian blinds, and open,
stop and close keys. The key of the direction the cover moves is filled; a
key that cannot move it further is disabled and stays out of the keys the
card's tick enables. Sliders show their value while dragged and send it on
release. The card offers only what supported_features allows. A
battery-powered blind shows its battery at the top right: the app sends
the device's battery sensor as the extra the vacuum card uses, and adds
current_tilt_position to the attributes.
Checking Studio 1 after its update to firmware 0.2.49 found that a setting
number equal to its initial state 0 (night brightness 0 %) was never
published, so Home Assistant kept it unknown and ESP Screens could not
change it; apply_screen_settings now publishes a number without a state.
The app also took "unavailable" for the device name during the restart
and wrote sensor.esp_screens_unavailable; diagnostics ignore unknown and
unavailable. The editor redraws its tiles, top bar and entity list when
the live stream opened a screen before the first full inventory (a busy
add-on), and loads app.js and style.css by content stamp after Safari kept
an old script.
Checked with 254 Python and 14 C++ tests, generate_packages --check, ESP32
builds of both profiles, and 20 API checks of the card on host builds of
both boards. Firmware 0.2.50.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A code check found that a setting changed on a screen could jump back.
ESP Screen Manager kept its own copy and sent it with every full update:
after a restart of the app or Home Assistant, every hour and after every
save. A change it missed while it or Home Assistant restarted came back
to the old value, a reported setting was lost when save() refused a
layout with a removed entity, and every step of a held - or + on the
screen resent the whole screen.
Firmware 0.2.49 offers every setting as an entity over the existing
preference records (no migration): Night mode, Night starts/ends,
24-hour clock, Back to page 1 and after how long, Back to page 1 on
standby, Swipe between pages and, on the Guition, Rotation, next to the
four numbers and Auto standby. The settings page and every entity write
through settings_screen::set(), which clamps like the page and stores,
applies and reports only a real change. screen_message answers a caller
that asks with its status and layout revision (api.respond). The
settings tile opens without Home Assistant.
The app reads and changes the settings of such a screen through its
entities and leaves them out of the layout message. Older firmware keeps
them with the layout; a setting it reports is stored without the tile
check and not sent back. Pings to screens that can answer learn at once
whether a tile is missing and resend after 30 s instead of 120 s. Layout
sensors come back after a Home Assistant restart, and weather tiles only
ask for the forecasts the entity supports. The editor's settings form is
now three cards with the rows of the screen's own page; a change applies
at once, and an offline screen shows its values as unknown.
Checked with 245 Python and 14 C++ tests, generate_packages --check,
ESP32 builds of both profiles, host-build API tests on both boards, and
a live run against Home Assistant 2026.9.1 with a temporary host device.
Firmware 0.2.49.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Since firmware 0.2.44 the back arrow of the brightness, colour and climate
cards barely responded, and so did the button at the top right of the
climate card. on_boot created the settings hold strip after the YAML cards
on home_page, so the transparent strip lay on top of them and LVGL's hit
test gave it the taps in the top bar region.
attach_hold now takes the first card as `below` and moves the strip and its
fill line to that index, under every card. Holding the top bar opens the
settings page from the tile pages only. The dim wake overlay was never
affected: apply_screen_settings moves it to the foreground when dimmed.
Checked with 215 Python and 14 C++ tests, generate_packages --check, and
ESP32 builds of both Easy profiles. Firmware 0.2.48.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A light's colour slider on Studio 1 sometimes sent red on release. Measured
live: an automation with a state trigger on a six-lamp group (no `to:`) set
"Standby after" to its current value after every colour change. The screen
reported the setting, ESP Screen Manager saved the layout and resent layout,
header and every tile state, and the screen drew its whole page again,
reading no touch input meanwhile. The GT911's stray (0, 0) contact then
became the finger's last position, and LVGL 9.5 set the slider from it on
release. Both red commits left 75-85 ms after such a resync.
The app no longer saves or resends a setting event whose value the layout
already has. The four setting numbers return before the preference write
and the event when nothing changed. A layout message that changes nothing
on the screen no longer runs layout_changed() and refresh_all(); the tile
states after it redraw their own tiles. The Guition wraps the LVGL touch
read callback with cyd::GhostTouch, which repeats the previous read for a
pressed (0, 0), and the edge swipe skips that sample too. Both slider
handlers log a jump to an end on release.
Checked with 214 Python and 14 C++ tests, generate_packages --check, and
ESP32 builds of both Easy profiles. Firmware 0.2.47.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
One rule ends a Sleep: a tap, Wake or an alert. The Auto standby switch
cleared sleep_requested when Home Assistant turned standby off, while
ESP Screens and the settings page on the screen did not; now only
wake_display clears the flag, and apply_screen_settings still wakes a
screen the standby time dimmed.
dim_display no longer checks anything itself (the interval checks Auto
standby, the Sleep button sets the flag). It dims, then runs go_home with
"also on standby" on, otherwise closes the settings page and every card
through close_cards. Before, it copied half of go_home and hid the four
YAML overlays itself, so a runtime detail card stayed open under a dimmed
screen. close_cards is now the one runtime_tiles::dismiss() call: the
fallback and the widget.hide actions after its return ran for nothing.
The backlight light becomes internal. Home Assistant could change it, but
the minute tick put the level back, and switching it off darkened the
screen without standby. The brightness numbers and Wake/Sleep replace it.
A dead clock-label branch from the manual profile leaves
apply_screen_settings.
Checked with 209 Python and 14 C++ tests, host builds of both boards over
the native API, and ESP32 builds of both Easy profiles. Firmware 0.2.46.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Every screen gets two template buttons, Wake and Sleep, for automations
(button.press on button.<screen>_wake / _sleep). Wake does what a tap does:
a dimmed screen runs wake_display, a screen that is on only restarts the
standby count. Sleep dims right away through dim_display, also with Auto
standby off, and closes an alert that is up (reported as remote); it is
ignored during a touch calibration.
A new non-restored flag, sleep_requested, keeps that standby in place:
dim_display dims with standby off, and apply_screen_settings, which runs
every minute, no longer wakes such a screen. wake_display clears it (a tap,
Wake, an alert), and so does the Auto standby switch when it really turns
standby off. Nothing is saved, so the buttons cost no flash writes and may
run on every motion.
The Claude skill, README, add-on docs and RELEASING describe both buttons.
Checked on host builds of both boards over the native API, 24 checks each
(timer, Sleep with Auto standby off across a minute tick, alert,
calibration), and with ESP32 builds of both Easy profiles (+336 B RAM).
Firmware 0.2.45.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Holding the top bar for about a second and a half opens a settings page on the
screen, with a blue line filling along the top edge while you hold. A built-in
screen.settings tile and the esphome.<screen>_open_settings action open it too.
The page is one constexpr table of rows in settings_screen.h: Brightness,
Night, Screen (clock, back to page 1, swiping, rotation) and This screen (name,
address, firmware, Home Assistant, Restart). Toggles flip on a tap, numbers and
times step with -/+ (a held time walks whole hours), a choice cycles in a chip,
and a row that depends on a switch above it is greyed out. The LVGL objects exist
only while the page is open: +608 bytes of RAM and +10.7 KB of flash on the CYD.
A change is saved, applied and reported through the existing screen_setting
event, so ESP Screens shows the same value within a second.
New setting auto_home (default on, 2 minutes): an open card, the settings page
or a later page goes back to page 1 after that long without a touch, not only on
standby. It travels beside the frozen eleven-key settings block with its own
preference, so older firmware and older apps keep working.
docs/SETTINGS.md is the recipe for adding a setting; the Claude skill explains
the page. Firmware 0.2.44.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Four Home Assistant events change a screen the way the app's own editor does:
esp_screens_add_tile (add or change), esp_screens_remove_tile,
esp_screens_move_tile and esp_screens_order_tiles. An event names the screen by
device name, shown name, area or title, and may set the size, the direct
control, the display, the icon, the colour and the spot; a tile with a control
or a forecast becomes double-width by itself. apply_tile_event returns the new
layout and Manager.save does the validating and the pushing, so nothing can get
past the twenty-tile limit, a duplicate entity or an occupied spot.
Every event is answered with esp_screens_tile_result, and every screen publishes
what it shows as sensor.esp_screens_<node>, with the page, row, column, width,
control and display per tile. So an assistant can read a screen before it moves
anything, and a refused event says why and changes nothing.
The Claude skill gains a Tiles section with the events, the fields, what each
kind of entity can do, how to read a screen first, and the rule to show the plan
and wait for a yes.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Xandrossian