ADR 0044's step 6b (D44, D46 to D48): a Viewer, and an agent, suggest
changes within a shared studyset, which its Owner and Editors accept or
decline.
A suggestion or report is written as Explore's are, in its writer's own
synced space, with kinds of its own (space-suggestion, space-report)
and the space's id in feedback_thread's publicationId, so no synced
table changes and older apps, skipping kinds they don't know, keep
syncing. Where the studyset is private it is sealed with its key, bound
to the space; a reply to it is sealed for its thread.
Learn's server gathers a space's threads for its members alone
(feedback.threadsInSpace, NotFound for anyone else): only threads and
replies by members now, decisions only from rows in the space itself,
where the engine lets its Owner and Editors alone write; sealed
payloads pass in base64 with their key and nonce. The inbox adds the
threads its caller decides on or wrote. The module tells a host a
person's spaces and each space's members (sharedSpacesOf,
membersOfSpaces). No migration: no table changes.
In the app, a Viewer's card and note offer Suggest a change: the fields
write into the suggestion alone, sent with a message. A shared
studyset's menu lists its Suggestions, where its Owner and Editors
accept a change as a revision, or a new card or note into the
studyset, and decide in the space. A suggestion naming a file that
isn't the studyset's, or bringing files of its own, isn't taken.
Whats-New: In a studyset shared with you as a viewer, suggest a change to a card or note; its owner and editors find your suggestions under Suggestions in its menu and accept or decline them.
Whats-New-De: In einem Lernset, das du nur ansehen darfst, schlägst du Änderungen an Karteikarten und Notizen vor; Besitzer:in und Bearbeiter:innen finden sie im Menü unter „Vorschläge“ und nehmen sie an oder lehnen sie ab.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
ADR 0044's step 6a (D42, D43, D45): accounts for programs, such as AI
assistants, that an identified person makes from the Profile.
halkyon_base keeps them on the instance's register, which its members
ask: an agent is an account (agent_account: its owner, its name, its
public keys, set by itself once) with one token (agent_token: 32
random bytes, kept as their SHA-256 alone, shown once). agents.create,
mine, newToken, revoke, delete and setKeys serve its owner;
agentSignIn.signIn exchanges the token for an ordinary server-side
session of a day, with the scopes identified and agent, at most 60
sign-ins an hour from a network. Revoking a token or making a new one
ends the agent's sessions; a token unused for 90 days lapses in every
service's pruning. Up to 5 agents a person, where the operator didn't
switch them off (AGENTS=off; said on the privacy page and in
instance.basics). An agent makes no agents, creates or owns no shared
space, is invited and raised as a Viewer alone, is never handed a
space on, is named no operator and granted no assistants; each refusal
has its test. Deleting a person deletes their agents first; an agent's
public profile names whose agent it is.
In the apps, every Profile has an Agents card (halkyon_app's
AgentsSection): made with a name and a handle, its token shown once to
copy, a new token, the token taken back, deleted. Learn's Share sheet
marks an agent among the members and offers it no role above Viewer.
The migrations, the module's and each host's (Learn, Meditate, Todo,
Train), are additive alone: CREATE TABLE "agent_account" with a unique
index on "agentId" (one row an agent account) and one on "ownerId" (a
person's agents, and deleting them with the person), and CREATE TABLE
"agent_token" with unique indexes on "agentId" (one token an agent)
and "tokenHash" (the sign-in finds the token by its hash). No existing
table is touched, so the server before the deploy keeps working on the
migrated database.
Whats-New: Make agents in your Profile: accounts for programs such as AI assistants, which sign in with a token and see only what you share with them.
Whats-New-De: Leg in deinem Profil Agenten an: Konten für Programme wie KI-Assistenten, die sich mit einem Token anmelden und nur sehen, was du mit ihnen teilst.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The apps' half of ADR 0044's step 5 (D37 to D40), on the instance's
half of the commit before.
halkyon_app's SharedSpaceGateway asks for inviting by mail, the
preview of a link, the code proving an address, accepting a mailed
invitation, a space's link (read, switched, joined) and marking a
share seen; SpaceInfo says whether a share is new and who made it,
a member whether they were invited by mail. joinTokenOf reads a link
or its token as pasted. The fake keeps the mails and codes it would
send, for the apps' tests.
In Learn:
- the Owner's Share sheet invites by handle or by email, lists an
invitation mailed by its role alone, and switches the space's link,
copying it while on;
- the Library's menu has Join with a link, and a link opened on
Android (/join/<token> on learn.halkyon.eu, verified through the
instance's assetlinks) opens the same on a screen of its own: who
invites and as whom, then Join, or for a mailed invitation the
address and the code mailed there;
- a studyset someone shared is new until opened: in the Inbox
("@alice shared Japanese with you", Open or Leave), on the Profile
tab's badge and as New on its library card (NewSharesCubit, read
again after every merge).
Whats-New: Invite people to a shared studyset by email or with a link, join one from the Library's Join with a link, and see in the Inbox what was shared with you.
Whats-New-De: Lade Leute per E-Mail oder Link zu einem geteilten Lernset ein, tritt einem über „Mit einem Link beitreten“ in der Bibliothek bei und sieh im Posteingang, was mit dir geteilt wurde.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Every route of a studyset says what the person may do there
(StudysetAccess, set in _inStudyset): whether it is shared with people,
and whether they only view and study it (Halkyon's ADR 0044, D35, D36).
A Viewer opens cards and notes on the same screens, read-only: the rich
fields show their content (LearnContentEditor shows LearnContentView),
the composer's plain fields are read-only and its changes go nowhere,
the note screen reads as a conflict leaves it, each with the line that
the studyset is theirs to view and study. The lists, the card browser
and a study session offer them nothing that writes: no new card, import,
edit, move, rename, reading deadline, reorder or delete. Their tags,
progress and reading stay theirs to change, being their own rows.
Burying and reporting a card are written on its row, which every member
shares, so they are hidden in shared studysets for everyone until they
become each person's own rows (D36).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
A passage was a Wrap of a Text per part between gaps. A Wrap places whole
children, so a part longer than a line took the width to itself, and every
gap and the text after it started a line of its own. One Text.rich with each
gap a WidgetSpan wraps the text as one paragraph, the gaps standing in it.
Whats-New: Fill-in-the-blank passages read as running text, their gaps inside the lines.
Whats-New-De: Lückentexte lesen sich als fortlaufender Text, die Lücken stehen in den Zeilen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
A search opened both sections whatever they were set to, so a tap on
their headers did nothing while one was under way. A search still opens
them, so their matches show, and a tap now folds them until the search
ends, which brings them back as they were left before it.
Whats-New: While searching the library, Archived and Recently deleted fold again when tapped.
Whats-New-De: Beim Suchen in der Bibliothek lassen sich Archiviert und Zuletzt gelöscht wieder einklappen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Share opens one sheet (Halkyon's ADR 0044, D29): the people first, then
the community. A studyset of the person's own offers Share with people,
which says what it does, moves it into a space of its own with its
progress told and Abort, and opens it there (StudysetSharingCubit). A
shared one lists its members (SpaceMembersCubit): the Owner invites by
handle as Editor or Viewer, changes roles, hands on and removes, and
deletes it for everyone; anyone leaves, the Owner naming who takes it
on; each refusal is said. A member's row opens their safety code
(twelve groups of five digits) to mark compared, and changed keys are
flagged, trusted only after comparing (D32). The community row leads to
the privacy dialog or publishing as Share did, and is off for a shared
studyset (D27).
A studyset knows its space and whether the person only views it
(Studyset.sharedIn, viewOnly, from the sync engine's grants). Its mark
is Lucide's users icon, "Shared · private" (D31). A Viewer's studyset
screen, its options, its topics and the move sheet offer nothing that
writes, and say they can view and study it (D30); the engine refuses
their writes on the device in any case. A shared top isn't moved,
archived or deleted from its options, and nothing moves in or out of
its space.
Whats-New: Share a studyset with people: invite them by their handle as editors or viewers, and compare safety codes to know nobody stands between you.
Whats-New-De: Teile ein Lernset mit Personen: Lade sie über ihr Handle als Bearbeiter:innen oder Betrachter:innen ein und vergleicht Sicherheitscodes, damit niemand zwischen euch steht.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Step 4 (e) of ADR 0044, its first part, for every app's screens.
- halkyon_app: `SharedSpaceGateway` on the module's `sharedSpace`
endpoint (the spaces, their members and roles, inviting, handing on,
leaving), refusals as `SpaceRefused`; `SharedKeys.trustIn` gives
each member's safety code and whether it was compared or their keys
changed, `markCompared`, and `trustNewKeys`, which pins the keys
whose code was compared and never those the instance hands next.
`FakeSharedSpaceServer` keeps the module's rules for tests, in step
with `FakeSharedKeysServer`, which now knows members' handles.
- Learn's sharing makes its space through it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Step 4 (d) of ADR 0044 (D14, D23, D27, D28), without its screens yet.
- `StudysetSharing`: refuses a studyset with a publication in its tree
and a device that can't seal, before anything is made; makes the
space, readies it, exports the tree with its progress, imports it
under new IDs with its top placed in the space, and deletes the old
one for good once the copy is whole. Failing or aborted, the space
is deleted and the studyset stays as it was.
- `SyncSpaceSetup`: syncs until the person's membership of the new
space came, then makes its first key and seals it to them.
- `Privacy` writes a top studyset placed in a space with that space's
key; a studyset can be inserted under an ID chosen beforehand.
- The archive's export and import take `moved`, keeping each item's
link to Explore and each card's origin.
- An import that fails midway takes back what it wrote, as an aborted
one did.
Whats-New: An import that fails partway no longer leaves part of the studyset behind.
Whats-New-De: Ein Import, der unterwegs scheitert, lässt kein halbes Lernset mehr zurück.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
ADR 0044, D8 and D9, its first step: before publishing to Explore, the
person confirms the notes, PDFs and other files that go. Cards and
learning paths always go.
Publish's Continue leads to a step listing them by kind, none ticked at
first but those the latest release held. Each kind is ticked whole, and
an item's menu ticks everything but it or nothing but it. What isn't
ticked goes to the server as publish's leftOut. So does whatever the
sync before publishing brings that the person never saw.
The server leaves those out of the release and its counts, and leaves
out the files only they name. leftOut is optional: released apps send
none and publish everything, as before. publish.releasedMaterial names
the notes and files of the latest release, and none for a studyset not
the caller's, as for one never published.
Whats-New: Before publishing to Explore, you choose which notes and files go with the cards and learning paths.
Whats-New-De: Vor dem Veröffentlichen auf Explore wählst du, welche Notizen und Dateien mit den Karteikarten und Lernpfaden mitgehen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
ADR 0043's recommendation, for Todo's attachments: what Learn had for its
files on the device moves into halkyon_app, unchanged but for its names,
and Learn uses it from there.
- FileCache, DirectoryFileCache (Learn's AssetFileCache): the files by
their SHA-256 in a folder of the person's, and which wait to be
uploaded.
- FileServerGateway, ServerpodFileServerGateway (Learn's
AssetServerGateway): halkyon_base's files endpoint, taking the module's
caller rather than Learn's client; FileLimitReached, FileRoom.
- Their fakes in halkyon_app's testing: MemoryFileCache,
FakeFileServerGateway, sha256Of.
- SealedFile: the format every app seals a file in (HLKF, version 1),
sealing with halkyon_app's ContentCipher; Learn's Privacy seals and
opens its files with it, for the same context as before, so every file
sealed so far opens as it did.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
- The Library's search lists titles starting with what is typed first,
then titles holding it as a word, then secondary titles, rather than
by the shelves' recency.
- The offer to make every studyset private is gone: each studyset's
lock already says and changes it, and new ones are private.
- Making a studyset private or not from its sheet flips the switch as
soon as the person agrees and says that everything in it is being
written anew, rather than standing still for a minute.
- Jellyfin's password can be shown while typed.
Whats-New: Searching the library lists title matches first; changing a studyset's privacy shows its progress; Jellyfin's password can be shown.
Whats-New-De: Die Suche in der Bibliothek zeigt Treffer im Titel zuerst; das Ändern der Privatsphäre eines Lernsets zeigt den Fortschritt; das Jellyfin-Passwort lässt sich anzeigen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Every read of the person's content refreshes their keys, and a refresh
asked the instance whether the account has keys until it once answered
yes, which an account signing in with SimpleLogin alone never does. So
on a poor connection every read waited on a request: the library's
progress, a request per studyset in a row, took over five minutes, and
Home was slow too.
A device holding the keys already knows the account has them, so the
keys now open first and the instance is asked only while no other sign
says so (a device without the keys still asks at every refresh).
Whats-New: The library and Home open quickly again without a good connection.
Whats-New-De: Die Bibliothek und die Startseite öffnen sich ohne gute Verbindung wieder schnell.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
A studyset's files take Markdown (.md, .markdown) as well as PDFs, in
the same synced rows with a kind (payload schema 3, written only for
Markdown, so older apps skip those rows). Upload a PDF becomes Upload a
file; a Markdown file must be UTF-8 text.
The reader opens one formatted and read-only through the editor's
Markdown reader, keeping fenced code, which that reader drops; the file
itself never changes, and its source text is a tap away, selectable too.
It has neither search, zoom nor screenshots. It reads in 100 parts by
its top-level blocks, so reading progress, deadlines and the cards and
notes tied to a place work as a PDF's pages do, shown as a share; its
outermost headings are its chapters.
Whats-New: Upload Markdown files beside your PDFs: they open formatted, with reading progress, deadlines and flashcards from what you select.
Whats-New-De: Lade Markdown-Dateien neben deine PDFs hoch: Sie öffnen sich formatiert, mit Lesefortschritt, Fristen und Karteikarten aus deiner Auswahl.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
A path shows either one step at a time (Steps) or on one page (Scroll,
as Tomorrow University's): the steps one below the other up to the first
question not answered right, answered in place with a lock below it, the
page running on once it is right, Finish at its end. A text step is done
once its end was on screen. The outline of the steps (done ticked, the
one in view marked, those beyond the question locked, a tap scrolling
there) is a column from expanded windows on, a drawer from the bar below.
The choice is the person's study option (`StudyOptions.pathView`,
settings schema 15), in Settings › Study and the session settings.
Whats-New: Learning paths can show on one scrolling page with an outline beside them: choose Scroll under Settings › Study.
Whats-New-De: Lernpfade lassen sich auf einer scrollbaren Seite mit Gliederung daneben zeigen: unter Einstellungen › Lernen „Scrollen“ wählen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The curator puts Tatoeba's readings on a dictionary's examples by their
Tatoeba ID (`withTatoebaReadings`, `curate readings`, and in `update`):
furigana over JMdict's Japanese from its jpn transcriptions, pinyin with
tone marks over Chinese from its cmn ones, each word over as many Han
characters as it has syllables. An example gets `ruby`, `[start, end,
reading]` each, and `rubyUnreviewed` where no one wrote or checked them
(MeCab's furigana, the automatic pinyin). JMdict's conversion goes to 3,
so instances convert again at once.
A word's example hints carry them as ruby marks, the word still in bold.
The hints sheet shows them from the start with a switch above
(Furigana, Pinyin), kept on the device (`DeviceSettings.hintReadings`),
and below a hint whose readings no one reviewed says so.
Whats-New: Example sentences in a word's hints now show furigana or pinyin over them, with a switch to hide them.
Whats-New-De: Beispielsätze in den Hinweisen eines Worts zeigen jetzt Furigana oder Pinyin darüber, mit einem Schalter zum Ausblenden.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
As Tomorrow University's challenges have it (the developer's decisions of
10 Nov 2026): beside a studyset's material, one at a time, its topics,
its introduction (the description, whole; left out where there is none)
and its bibliography, read and changed in place. In the side column from
expanded windows on, in the topics' sheet below; it opens on Topics, at
every level. The description no longer stands above the material, nor
the bibliography's button at the foot of the topics. The pill between
the material's parts becomes PillSwitch, shared by both.
Whats-New: A studyset's topics, introduction and bibliography now share one panel beside its material, a tap apart.
Whats-New-De: Themen, Einführung und Literatur eines Lernsets teilen sich jetzt ein Feld neben dem Material, einen Tipp voneinander entfernt.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The library as Vaia's (the developer's decisions of 10 Nov 2026): the
starred on a shelf of their own, then those studied or changed in the
last 7 days, the last month and before, each under its heading, the
archive last. Recent is the later of a card's last rating
(StudyProgress.lastStudied) and the studyset's own last change. A
dropdown narrows the library to one of the subjects in use, kept on the
device (DeviceSettings.librarySubject). A card names its subject under
its title, with its icon, and a language course its language beside.
Whats-New: The library groups your studysets by when you last used them, and narrows them to one subject.
Whats-New-De: Die Bibliothek ordnet deine Lernsets danach, wann du sie zuletzt genutzt hast, und zeigt auf Wunsch nur ein Fach.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
A top studyset is filed under one of 44 subjects in six groups (the
developer's decisions of 9 and 10 Nov 2026): languages as one, computing
in five, Other / not sure last. Each has a Lucide icon and a default
colour, which picking it gives the studyset, its own to change from then
on. The studyset's sheet picks it (pickSubject); a sub-studyset shows its
top studyset's (subjectOf) and offers none. The payload keeps it as
`subject` (schema 8), so archives and published studysets carry it; an
ID this version doesn't know is kept and reads as none.
Whats-New: A studyset can now be filed under a subject, which gives it that subject's colour.
Whats-New-De: Ein Lernset lässt sich jetzt einem Fach zuordnen, das ihm dessen Farbe gibt.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Exporting a studyset said "Exporting…" in a snack bar until the place to
keep it was asked. It now shows a sheet with its steps (reading the
content, gathering files for an archive, packing), how many of the step's
items are done and the percent of all of it, and Abort, which keeps
nothing (decided 9 Nov 2026); then asks where to keep the file as before.
Both exporters check between studysets, and the archive's between files;
packing runs on its isolate to its end, its result dropped once aborted.
ExportCubit runs an export's work behind the sheet (runExport), which
says how it ended: made, failed (a file missing named as before) or
aborted.
Whats-New: Exporting shows its steps and how far it has come, and Abort stops it.
Whats-New-De: Der Export zeigt seine Schritte und wie weit er ist, und „Abbrechen“ beendet ihn.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Importing an Anki deck showed one bar and "Importing…". It now says which
of its four steps it is in (copying files, writing flashcards, tagging,
bringing over the review history), how many of the step's items are done
and the percent of all of it, and offers Abort (decided 9 Nov 2026), as
the archive's import does.
The importer checks between files and batches of cards. Aborted, it
deletes what it wrote: a studyset it made for the deck for good with all
of it; importing into an existing studyset, the cards it wrote there and
the topics it made for subdecks, leaving what the studyset held before;
and the files' rows it made, not those the person had. Tags it made stay.
Once everything is written, Abort no longer undoes it.
Whats-New: Importing an Anki deck shows its steps and how far it has come, and Abort stops it, keeping nothing of it.
Whats-New-De: Der Import eines Anki-Decks zeigt seine Schritte und wie weit er ist, und „Abbrechen“ beendet ihn, ohne etwas davon zu behalten.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Importing a Halkyon archive showed one bar. It now says which of its three
steps it is in (copying files; writing studysets, flashcards and notes;
bringing over the progress), how many of the step's items are done and the
percent of all of it, and offers Abort (decided 9 Nov 2026).
The copier checks between items, never leaving one half written, and
keeps a trail of what it wrote: the top studyset once made, and the files'
rows it made (not those the person had, found before each file is added).
Aborted, the importer deletes the top studyset for good with everything
inside it, as Delete now does (the reviews, answers and pages read going
with their cards, steps and PDFs), and those rows; the tags it made stay,
as names the person may well use anyway. Once all of it is written, Abort
no longer undoes it.
WorkProgress, WorkStages and AbortSignal (core/work/) and
WorkProgressView (core/widget/) are shared with the Anki import and the
exports, which follow.
Whats-New: Importing an archive shows its steps and how far it has come, and Abort stops it, keeping nothing of it.
Whats-New-De: Der Import eines Archivs zeigt seine Schritte und wie weit er ist, und „Abbrechen“ beendet ihn, ohne etwas davon zu behalten.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The tab saying that its material is in the studyset's topics stood at the
top of the room below the tabs, while an empty tab's message stands in
its middle (SliverCentered), so switching between them moved the message
up and down (the developer's review, 9 Nov 2026). It stands in the middle
too now.
Whats-New: A studyset's tab whose material is in its topics now says so in the middle, as empty tabs do.
Whats-New-De: Ein Tab eines Lernsets, dessen Material in seinen Themen liegt, sagt das jetzt in der Mitte, wie leere Tabs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The books of the old app's library run past the 50 MB a PDF could have
(decided 2026-10-03); the server and the app take up to 100 MB now. Other
files keep their own, smaller limit.
Whats-New: PDFs can now be up to 100 MB.
Whats-New-De: PDFs dürfen jetzt bis zu 100 MB groß sein.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
On Android the microphone now opens the app handling
RECOGNIZE_SPEECH (FUTO Voice Input, Google's), the default the person
chose in Android, and puts the text it returns where the cursor stood,
as Dictation does; cancelled, nothing. Only where none is installed, or
it can't be opened, does speech_to_text listen in line as before.
halkyon_app: VoiceInputApp (one listen, an answer once closed) on the
channel app.halkyon/voice_input, which DeviceSpeechInput asks first at
every listen, answering SpeechStart.finished so that Dictation doesn't
listen again; speech_to_text isn't touched on that route, so our app
asks for no microphone. SpeechInput.listen and Dictation.start take the
prompt the voice input app shows (the field's hint).
Each app: VoiceInput.kt (queryIntentActivities, startActivityForResult,
the first of EXTRA_RESULTS) attached in MainActivity, which hands it
activity results, and RECOGNIZE_SPEECH among the manifest's queries.
The privacy pages of Learn and Meditate say so.
Our own Whisper and Meditate's dream recordings are dropped (D18).
Whats-New: The microphone now dictates through your phone's voice input app, such as FUTO Voice Input.
Whats-New-De: Das Mikrofon diktiert jetzt über die Spracheingabe-App deines Telefons, etwa FUTO Voice Input.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Rating a card while studying waited for its review to be written, then
for the card's schedule to be replayed, before the next card showed.
StudySessionBloc now starts keeping the review and shows what comes
next at once: the card's new place in the round is worked out from its
reviews with the new one counted as if kept already
(ReviewRepository.schedulesAsStudied's pending, each review counted
once should it be kept by then).
Should keeping it fail, the step leaves the session's Undo, the rating
leaves its accuracy, and the card comes back where it stood, saying so
(withRatingFailed), as a failed rating did before. An Undo of a rating
still being kept waits for it and takes it back once kept; one that
couldn't be kept has nothing to take back, and the Undo has brought
the card back already.
Whats-New: While studying, the next flashcard shows the moment you rate one.
Whats-New-De: Beim Lernen kommt die nächste Karteikarte, sobald du eine bewertet hast.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Moving the chosen cards to another studyset wrote them one at a time,
each looking up its privacy keys and writing its row, while the
browser still showed them. CardBrowserBloc now shows the move at once
(gone from here, or after the cards of the studyset they go into where
the browser shows it, _withMoves) and puts them back where the move
fails, counting the failure for the screen's snack bar
(moveFailures, cardBrowserMoveFailed).
CardRepository.moveTo hands the gateway every card's new position at
once (CardGateway.updateStudysets): SyncCardGateway reads them in one
query, checks every card's key before changing any (one private card
among them refuses the whole move, as before for each), rekeys those
that need it, and updates every row in one write and one sync request.
updateStudyset delegates to it.
Whats-New: Flashcards moved to another studyset in the card browser move at once.
Whats-New-De: Im Karteikarten-Browser in ein anderes Lernset verschobene Karteikarten sind sofort dort.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
In a discussion, accepting an answer (or none), hiding or showing a post
and deleting one waited for the instance with every action disabled
(busy). DiscussionBloc now shows the discussion as the action leaves it
at once (Discussion.withAccepted, withPost, DiscussionPost.copyWith), the
actions staying available, and puts it back as it was where the instance
refuses (the refusal's snack bar as before) or can't be reached, then
reads it anew. Answering, commenting and editing still wait, as what
the instance makes of them is theirs to say.
In the feedback inbox, a thread opened shows read at once, a failure to
record it said to nobody (the next read shows the truth), and a decision
(closing, reopening, hiding, showing, declining) shows at once and is
put back with the failure's snack bar where it isn't written
(FeedbackThread.seen, decided). Accepting a suggestion still waits, as
it changes the author's content and can fail with SuggestionNotTaken.
Whats-New: Accepting, hiding and deleting in discussions, and closing or declining feedback, show at once.
Whats-New-De: Annehmen, Ausblenden und Löschen in Diskussionen sowie Schließen oder Ablehnen von Rückmeldungen zeigen sich sofort.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Deleting a kept conversation asked first and then waited while its
turns were blanked and deleted. It now goes as a card or a note does:
ConversationDeletions, a DelayedDeletions opened with the person's
data, takes it off the history at once (the list leaves out what is
pending), offers Undo for a moment on the sheet (hasSnackBars), and
blanks and deletes it once that is over; a deletion that fails shows
the conversation again, as every DelayedDeletions does. The question
before deleting goes, as Undo replaces it.
Whats-New: A conversation with an assistant is deleted at once, and Undo brings it back.
Whats-New-De: Eine Unterhaltung mit einem Assistenten wird sofort gelöscht, und „Rückgängig“ holt sie zurück.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Deleting a studyset for good blanks and deletes every card, note, path
and PDF of its subtree before the studyset's row goes, which took about
3 s while the studyset stayed in Recently deleted. Now the repository's
Recently deleted leaves it out at once (StudysetRepository.leaving and
stay, applied to the value last read without reading again), the
deletion runs in the background, still one at a time in
StudysetDeletions' queue, and where it fails the studyset shows there
again, read anew with what is left of it, and a snack bar says so.
Whats-New: Delete now in Recently deleted takes a studyset away at once.
Whats-New-De: „Jetzt löschen“ unter „Zuletzt gelöscht“ entfernt ein Lernset sofort.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer decided on 8 Nov 2026, after trying the variants on the
phone: the wave in the colour of the screen in view, without Material's
ripple, the indicator kept and there at once, a 1.2 s wave, top corners
of 20 dp, the bar's own neutral surface. Learn's bar stays cobalt, the
colour its tabs wash in, rather than its icon's purple.
So the bar has one look: FlushNavigationBar.waveDuration and
cornerRadius are constants, and the debug section goes from the four
Settings screens with what it switched (NavigationBarDebugSection,
debugNavigationBarStyle, effectiveNavigationBarStyle, the style record
and the variants A, C and F, the speeds, the corners and the tinted
surface). Their tests go too; those stating the look kept stay. Learn's
sign-out helper no longer has to scroll past the section.
ADR 0030's amendment of D60, the halkyon_ui README and Learn's
AGENTS.md record the decision and the alternatives set aside.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer's review of 8 Nov 2026: the bar follows the screen in view
rather than one colour per app, so a Todo project of its own colour gives
the bar that colour while it shows. The wave, the chosen destination's
icon and label and its indicator take ScreenTint.color, the colour the
screen in view claims through its TopGradient; the frame's accent, the
app's colour, stays only for the moment before any screen claims one,
since ScreenTint itself starts at cobalt, which would be wrong for Todo,
Train and Meditate. Learn's tabs claim cobalt, their wash, so its bar is
cobalt there now, not purple.
A colour changing on the same tab (a project opening) passes into the
next over 300 ms, at once with animations off. A wave paints in that
colour as it goes: it starts in the colour of the screen being left, as
the arriving one claims its own only once it shows, a frame later, and
passes into it with the rest of the bar.
A person's colours may be light, so the chosen destination's colour is
made legible (readableTone, halkyon_ui): a palette colour starts from its
own text tone as Todo's dates take it (its outline in light mode, its
fill in dark), any other colour from itself, then darkens towards black
in light mode and lightens towards white in dark, only until it reads at
4.5:1 on the bar and on its indicator. Every palette colour is tested in
both themes, on the neutral and the tinted surface. This also lifts
Learn's label, which read at about 3:1.
The debug section gains Bar surface, neutral or tinted: tinted mixes 6 %
of the screen's colour into the bar's own surface in light mode and 8 %
in dark, changing with it. Neutral by default, and in the shipping style.
Whats-New: The navigation bar takes the colour of the screen you're on, such as a project's own.
Whats-New-De: Die Navigationsleiste nimmt die Farbe des Bildschirms an, auf dem du bist, etwa die eines Projekts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Todo stores editor documents too, so the document, the editor, its view,
the format bar, links, the clipboard, scripts and EditorBlocks move, with
their strings and tests, to a shared package on the same appflowy_editor
fork. What is Learn's alone stays in Learn on top of it, and what the
editor took from it directly is inverted: EditorBlocks gains textMarks,
formatActions and onEditorClosed, RichContentEditor a wrapEditor, and an
InsertableBlock's name takes the BuildContext. Learn's screens use
LearnContentEditor and LearnContentView, which add Learn's blocks and the
grammar check. Learn behaves as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer's review of 4 Nov 2026. "Japanese on this device" becomes
"Dictionaries on this device", with one row a language: its name at the
start, what the device holds or that the download failed beneath it, and
its action at the end (Download with its size, the progress, Remove),
built as a list of rows so Korean's and Chinese's line up beneath
later. Japanese's is the only one now. The help speaks of a language's
dictionaries in general, then of what Japanese's hold (MeCab's, JMdict).
Whats-New: Settings lists the dictionaries on this device by language, each with its own Download or Remove.
Whats-New-De: Die Einstellungen listen die Wörterbücher auf diesem Gerät nach Sprache, jede mit eigenem Herunterladen oder Entfernen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Arranging Home kept no order of its own: a drop showed the old order
until the layout saved came back through the settings, once the
dashboard's changes settled 300 ms later, so the widget snapped back and
then jumped. _Arranging now holds the layout as arranged, shows each
change at once and saves it behind, taking a layout from the settings
only once its own changes are back (or, with none pending, as changed
on another device). The handle was an IconButton with an empty
onPressed inside the drag listener, whose tap and tooltip competed with
the drag; it is a plain icon now, named for screen readers.
Whats-New: Arranging Home's widgets, a widget stays where you drop it, and dragging starts at once.
Whats-New-De: Beim Anordnen der Widgets auf der Startseite bleibt ein Widget, wo du es ablegst, und das Ziehen beginnt sofort.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer's review of 4 Nov 2026. The entry goes from a studyset's
Add material, and with it the sheet for pasting a text
(proposeCardsFromText, _TextSheet), its three texts and proposeCards'
fromStudyset, which only it set to false. A note's and a PDF page's
flashcards stay; their tests now drive the proposals sheet from a note.
Whats-New: Add material no longer offers Flashcards from a text; a note's or a PDF page's flashcards stay.
Whats-New-De: Material hinzufügen bietet keine Karteikarten aus einem Text mehr an; die aus einer Notiz oder einer PDF-Seite bleiben.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Inside a studyset the field flickered as it opened and closed: the
floating location switched between a raised centre and the centre,
which Scaffold animates by scaling the button out and in. It keeps one
location now, raised above the licence line by padding. While open it
no longer slides away as the studyset or the Library scrolls
(ScrollHidingFab.pinned, new), and the keyboard covers Home, the
Library, a studyset and a session rather than shrinking them; the field
lifts itself by as much of the keyboard as reaches above it.
NavigationFrame no longer resizes for the keyboard itself, leaving it
to each tab's own scaffold, which still does by default (Todo's,
Train's and Meditate's tests pass unchanged).
Whats-New: Inside a studyset, the question field opens without flickering, stays while you scroll, and the screen no longer jumps when the keyboard comes.
Whats-New-De: In einem Lernset öffnet sich das Fragefeld ohne Flackern, bleibt beim Scrollen stehen, und der Bildschirm springt nicht mehr, wenn die Tastatur kommt.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Each screen's button kept its own open state and its own draft, so the
question half written on the Library was gone on Home. ComposerCubit,
provided beside ConversationDockCubit, now holds whether the field is
open and the draft (text, files attached, parts chosen) for every
button. Switching tabs keeps it as it is (AppShell says so before
goBranch); going to another screen within a tab, or over the tabs, keeps
it open only while something is written or attached. Only the button of
the screen at the router's location unfolds, so no tab kept alive
behind holds a second field.
Whats-New: The question you are writing stays open as you switch tabs.
Whats-New-De: Die Frage, die du schreibst, bleibt offen, wenn du den Tab wechselst.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer's review of 4 Nov 2026. Below the question, × and + stand
at the start, who answers is centred and the conversations and the
microphone stand at the end, every button the same rounded square, the
microphone listening and Send too. The row above the text ("Only what I
choose · 2 parts") goes, as +'s sheet offers the same choice; a web
search gets + with that row alone. Who answers shows its model's name
alone ('Haiku 4.5', AssistantProvider.modelName: Anthropic's without
"Claude", others as they are), outlined in its provider's colour (D60).
The microphone is halkyon_app's Dictation: what is said goes where the
cursor stands, and it listens again after each pause until stopped. It
gives way to Send once anything could be sent, as Claude's field does.
Whats-New: The question's buttons are in one row: close and + on the left, who answers in the middle, Conversations and the microphone on the right.
Whats-New: What goes along is chosen under +, for the web search too.
Whats-New: Who answers shows the model's name, in its provider's colour.
Whats-New: The microphone keeps listening after a pause, writing where the cursor is, until you stop it.
Whats-New-De: Die Knöpfe der Frage stehen in einer Reihe: Schließen und + links, wer antwortet in der Mitte, Unterhaltungen und Mikrofon rechts.
Whats-New-De: Was mitgeht, wählst du unter +, auch für die Websuche.
Whats-New-De: Wer antwortet, zeigt den Namen des Modells, in der Farbe seines Anbieters.
Whats-New-De: Das Mikrofon hört nach einer Pause weiter zu und schreibt, wo der Cursor steht, bis du es stoppst.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
On GrapheneOS, which installs no recogniser, the microphone did nothing
and stayed listening: speech_to_text's initialize throws a
PlatformException (recognizerNotAvailable) there rather than answering
false, and DeviceSpeechInput let it through. A listen now answers how it
went (SpeechStart: listening, noRecognizer, unavailable) and never
throws; Learn's field and Meditate's dream say that no speech
recognition is installed and name two to install (FUTO Voice Input,
Google's speech services), in halkyon_app's texts. It also passes
androidNoBluetooth, as the apps declare no Bluetooth permission the
plugin would otherwise ask for.
Whats-New: Where the device has no speech recognition, the microphone says so and how to get one.
Whats-New-De: Wo das Gerät keine Spracherkennung hat, sagt das Mikrofon das und wie du eine bekommst.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer's review of 4 Nov 2026: "an actual conversation with the AI
chatbot is not desirable and necessary". The field's voice-mode button,
the screen it opened, reading answers aloud and its texts go, and with
them flutter_tts, its patched copy in third_party/ (one override fewer),
the TTS_SERVICE query in the manifest and the plugin's registrants. The
microphone stays and dictates into the question. ImageApproval is the
conversation sheet's own again, as only voice mode needed it.
Whats-New: Voice mode is gone; the microphone dictates into your question.
Whats-New-De: Der Sprachmodus ist weg; das Mikrofon diktiert in deine Frage.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer's review of 3 Nov 2026: a studyset is made private, or
non-private, in Edit alone. The lock's sheet still says what its privacy
means, offers publishing one that isn't private, and finishes one only
partly private.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
What explains a setting or a section, rather than what is the person's
own, moves behind halkyon_ui's HelpButton, beside its title through the
new ExplainedTitle (the developer's review of 3 Nov 2026): in Learn's
settings (citations, the Pomodoro timer, grammar, the web search,
Jellyfin, the device's dictionaries, counting one's learning), the study
settings, export's progress, the Jellyfin tags and the study button's
back first; in halkyon_app, setting up a password, adding SimpleLogin,
own keys not allowed, the profile page's link back and the operator's
grant sheet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Beside the field's microphone, a button opens voice mode over the whole
screen: a glow rising from below, a big microphone, and +, who answers,
the speaker and End in one row. It listens with the device's recogniser,
asks what was heard in one kept conversation, reads the answer aloud
with the device's speech (flutter_tts, its Markdown left out) and
listens again. On Linux there is none, as there is no microphone; where
nothing reads aloud, answers are shown as text.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Following the developer's review of 3 Nov 2026 (ADR 0021, D67):
- + below the assistant's field, and below a conversation's follow-up,
opens "Add to your question" as Claude's "Add to session": three big
tiles side by side, Camera (where the device takes photos) and Photos
for an assistant reading images, and Files; below them, in the field, a
row leading to what goes along, showing how far the assistant looks and
how many parts are chosen.
- Camera and Photos through image_picker (flutter.dev's own): the camera's
own app and the system's photo picker, no permission asked on Android, a
file dialog on the desktop, which has no Camera; photos made at most
2048 pixels a side, read on the device like any file attached.
- What goes along is chosen in rounded rows, as Claude's Permission, in
place of radio buttons and chips: how far it looks with a tick, private
studysets with a switch, the parts each with a tick, a card's image with
its thumbnail.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The developer's review of 3 Nov 2026: Make private and Make non-private
leave a studyset's options (its card's menu, its screen's); a top-level
studyset's Edit has a Visibility section instead, its switch making it
private, or non-private, at once once the person agreed in the dialog as
before (ADR 0019), not on Save, and what that means behind the question
mark. A topic's Edit offers none, its privacy being its tree's. The lock on
a studyset's card still explains its privacy and keeps its own button.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Following the developer's review of 3 Nov 2026 (ADR 0021, D67):
- Unfolded in the Library, the field spans the screen between its margins,
as on Home and a studyset: it fell short on the right, by the padding
around the two folded buttons, which now stands only around those.
- Who answers stands right before the buttons at the right, which no longer
shift as its name changes (Claude Haiku, Claude Opus): it fills the space
left, aligned to its end, rather than a share of it.
- Every button of the field and of a conversation's follow-up is boxed, a
tile a shade above the field with rounded corners, as Claude's apps have
them; who answers too.
- On a studyset with a licence, the unfolded field takes its usual place
over the licence's line rather than squeezing in above it; folded, its
row still keeps clear of the line.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Learn's `SpeechInput` and `DeviceSpeechInput` (ADR 0021, D67) are
halkyon_app's now, and its `FakeSpeechInput` is in `testing.dart`, so
Meditate's dream journal (ADR 0038, D15) listens through the same code.
Learn imports them from there and behaves as before: its composer and
its tests are unchanged but for the imports, and `DeviceSpeechInput()`
asks for what it asked before.
`DeviceSpeechInput` takes three options, off by default:
- `preferOnDevice`: on Android, recognition on the device where it can
(Android 12 on), else offline; should that fail before a word is
heard (not a pause), it listens as Android otherwise does from then
on.
- `pauseFor`: how long a pause may last before the recogniser stops.
- `listenFor`: how long one listen may last at most.
`Dictation` writes what is heard where the cursor stands
(`DictationInsertion`: in place of a selection, set apart by a space),
shows the words so far, and listens again after each pause while it is
on, stopping by itself after three listens running heard nothing.
speech_to_text moves from Learn's pubspec to halkyon_app's, the lock
unchanged. Todo, Train and Meditate link the plugin now; their
generated plugin registrants (macOS, Windows) say so. An app declares
the microphone itself, and Todo and Train declare none, so nothing
there can listen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
The public profile was already one per person for every service of an
instance (the register, ADR 0016); its app side was Learn's alone. It
moves into halkyon_app: the model, the repository on the module's
endpoints, the form, the avatar drawn on the device and the sheet with
someone's profile. They take the Public prefix (PublicProfileRepository,
BasePublicProfileGateway), as Train's own ProfileRepository holds the
body's figures. AppDependencies gains profiles and files, the picker the
avatar is chosen with, which moves out of Learn too.
HalkyonProfile is every app's Profile screen: the person's own tab first,
the same in every app (the account and the ways they sign in, the app's
own cards, then the public profile), then the app's own tabs, folding into
the title. Learn adds its inbox and Statistics. Train adds Progress, Days
and, where nothing records, Plans, and so opens on the person's tab, no
longer on Progress. Todo, which has no Profile tab, opens it from a row at
the top of Browse, where Todoist keeps the account.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
Home gets a widget of its own, Pick up learning, below the goal: the
studysets at the top of the library that were studied, the one rated last
first, so one studied long ago sinks. Three show, the others behind Show
more. Each shows its name, which opens it, how much is learnt, when it was
last studied, and today's task (cards to review, new cards to learn), which
starts studying it, outlined while reviews wait. The progress is the
library's, worked out once for every screen; when each was last studied
comes from the reviews Home reads anyway. With nothing studied it takes no
room.
A dashboard arranged before it gets the widget below the goal, wherever
the goal is: a widget a layout misses now goes after the one before it in
a new dashboard, no longer at the end.
On a studyset's Progress, the tiles are filled without an outline, To
review has a history icon (amber while cards are due), the learnt bar is
thicker, and the exam follows the figures, as Vaia asks for its date
there: an invitation to choose it, opening the study plan's sheet, or,
once chosen, how ready the person is.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
