`UsageDelta` gains `prefillMs`, llama-server's own `timings.prompt_ms`, so the
footer under a finished reply is "read 9.5s · 50.3 tok/s · 3:00 PM". Prefill is
the half of a turn that was invisible and is often the larger: measured on the
0.6B here, 1m 4s for the first turn after a model loads against 22ms for the
next, whose prompt the server still had cached.
The clock moves to the end of the line. Everything in front of it is a
provider's own measurement, so a session on another provider has fewer of them
or none, and a reader who has learned where the time is should not have to find
it again because the model changed. The costs grow leftwards into the space
instead, and a test asserts every shape of the line ends with the same thing.
Verified on the emulator against a real llama session: three replies reading
"read 1m 4s · 193 tok/s · 3:54 PM", "read 25ms · 308 tok/s · 3:54 PM" and
"read 22ms · 194 tok/s · 3:54 PM", with the clock in one column.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A turn has two waits in front of the first token and they were one word.
`SessionStatus::Loading` was already the model coming off disk; this adds
`SessionStatus::Reading` for llama-server processing the prompt -- emitted when
the request goes out, cleared by the first thing the model says of any kind, so
it covers every generate in a tool loop rather than only the first.
Prefill is the expensive half on this machine: measured 9.5s for 6,068 tokens
and 22s for 14,068 on the 27B with the GPU to itself. Reported as `running`
that was indistinguishable from a model thinking, which is the thing the reader
is waiting for. The phone draws both with the working spinner and its own
words -- "loading model" and "reading prompt" -- and the session screen's
status row now spins for all three busy states instead of only `running`,
which is also how `loading` stops being a bare word with nothing moving.
Measured while checking the tok/s figure, and recorded in the rigs skill: the
27B holds 55.5 to 50.3 tok/s between 1.5k and 14k of context, so decode decays
gently, while the 0.6B on the CPU falls 30.1 to 11.5 over 6k. A shared GPU is a
different failure -- the model does not load at all.
Verified on the emulator against a real llama session: "loading model" while
the server started, then "reading prompt" with the spinner through prompt
processing, then the thinking card.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A llama.cpp session's `reasoning_content` becomes `Event::Thinking` deltas
closed by an `Event::ThinkingDone` carrying the span the driver measured, and
the phone draws it as a card of its own: "Thinking" with the spinner a running
command has, then "Thought for 12.4s". Deliberately not a tool call, so a run
of calls cannot collapse the reasoning into "Called 6 tools"; the reasoning is
also kept out of the next prompt, which `conversation` already ignored.
`UsageDelta` gains `tokensPerSecond`, the provider's own figure or nothing --
llama.cpp reports `timings.predicted_per_second` and the coding CLIs report no
such thing -- and a finished reply carries a small line under it saying when it
was sent and, where there is one, how fast it came out: "3:00 PM · 149 tok/s".
The compact usage bar drops the provider's name for the window and puts its
length after the time left instead: "42% · 3h 20m left / 5h".
Three things that had to come with it: the transcript coalesces runs of
thinking deltas as it does reply deltas, so one block is one row of a page
rather than a page of its own; `joinPages` welds a block cut by a page boundary
(`healSplitThinking`), since the half with no ending spun for ever; and
`UsageDelta` now reaches the fold, which is what carries the rate to the reply.
Verified on the emulator against a real Qwen3-0.6B session and the echo rig's
new `/think [seconds]`: the spinner while it runs, "Thought for 1.4s" and
"2:54 PM · 149 tok/s" after, the reasoning on tapping the card, and the usage
bar reading "42% · 3h 19m left / 5h".
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Two things a session could not say, and one it was saying wrongly.
**Every provider setting is reachable.** `-np 1`, the MTP draft depth, the
tool set, the sampling parameters -- most were hardcoded to what measured
best on this machine, which is right as a default and wrong as a constant:
the next machine has a different GPU and a different core count, and
nobody running this app can edit the source. `DriverKind::params` now
declares what a provider takes -- key, label, shape, what blank means, and
whether a change waits for a restart -- and the phone renders whatever
arrives, on the spawn form and in the session settings dialog. Adding a
setting to a driver is one entry in that table and no app change.
`POST /sessions/{id}/params` takes the whole map, so an absent key is the
instruction to unset; the sampling half applies at once and the session is
told in words which of the rest are waiting for a restart.
`tools` is one of them, because it is the biggest lever on a tight
context: the seven built-in definitions are ~1,300 tokens of every prompt
(2,191 against 887 with none). `"none"` omits the flag rather than passing
it on, since `--tools none` is `unknown tool "none"` and a server that
exits.
**The context figure has a denominator.** `Event::ContextWindow` carries
it, read from `llama-server`'s `/props` once the model is up -- the
measurement rather than the request, since a session that named no context
size gets the model's own. Neither coding CLI states its window, so those
keep the bare figure: "2,042" and "2,042 / 8,192" are deliberately
different-looking, and a missing ceiling is never drawn as a proportion of
an assumed one.
**And the numerator was wrong**, by the length of the last reply: it was
the prompt alone, so a five-word answer reported 2,042 against a slot
holding 2,355. It is the turn's total now, which matches `llama-server`'s
own `n_tokens` to within a token.
Two defects the review found, both of which would have shipped: changing
settings on a *stopped* session reported "no process running, so it can't
take new settings", when a stopped session is exactly when you would set
them for the next start; and `GET /tools` answers **403** rather than an
empty list on a server started without `--tools`, so reading it as a
failure made the no-tools session one that never started.
Verified against real models: settings spawned and changed live, the
restart note, a session with two tools and one with none, and the counter
checked against the server's own slot occupancy each time.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A llama session was a chat box: no tools, a fixed model, no permission
mode, and a model name drawn as the path the file sits at. It now runs the
agent loop itself, which is what the pieces below all hang off.
Tools are `llama-server`'s own (`--tools all`), which that server both
publishes and runs -- `GET /tools` for the definitions, `POST /tools` to
call one. Web search is Exa's MCP server, reached from this backend rather
than from the machine serving the model: that is what llama.cpp's own web
UI does, and it puts the search on the machine with a route out instead of
the one with the GPU. `llama-server`'s `--mcp-servers-json` can only spawn
local commands, so using it would have meant a Node bridge on every
machine that serves a model.
Driving the loop is what makes the permission gate ours. Two modes,
`manual` and `bypassPermissions`, which is what the mechanism has: the web
UI asks before every call and remembers the tools you say "always" to. The
allowances fold back out of the transcript's own answers, so they survive
a restart and a model change without being stored anywhere else.
Also here, because tools made each of them matter:
- **Loading is a state.** A 12 GB model takes twenty seconds to reach
memory and refuses everything until it has; the session used to report
`running` for that whole time, and a message sent meanwhile came back as
an error. It is `loading` now, and the message waits.
- **The model can be changed.** A `llama-server` holds one model, so this
stops it and starts another. The conversation survives because it was
never in the server.
- **Models are named, not pathed.** `general.name` read out of the file
itself -- over ssh too, in the round trip the spawn was already making.
Where two models share a name the file name breaks the tie.
- **`-np 1`, and the MTP draft head where the file has one.** Measured on
the 27B here: 41.5 tok/s plain, 61.4 with `--spec-type draft-mtp` at one
slot, and 28 with it at four -- speculating against a split KV cache is
worse than not speculating. The flag is conditional because asking for a
head that is not there makes `llama-server` exit.
- **A refusal says what to do.** Tool results are thousands of tokens, so
an overrun context is now ordinary; it was "http status: 400" and is now
the server's own "exceeds the available context size, try increasing it".
`GET /machines/{id}/models` is gone: the provider models route answers the
same question, and two answers to one question is how a picker comes to
offer a model the spawn screen does not.
Verified end to end against real models: a tool call asked and allowed, an
Exa search, a shell command, a 27B loaded while a message waited on it, a
model switch mid-session, a second message queued behind a running turn,
and the whole of it again on a session running over ssh.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Removing `Event::TaskNote` hours after adding it made every transcript that
had recorded one unreadable. `Transcript::open` parses every line, so `launch`
failed for those sessions and `SessionManager::new` logged
"couldn't relaunch session <id>" and skipped them -- and a skipped session has
no pump and no driver. On the phone that is no status, no history and nothing
sendable, for every live session that had run a background task. One
unfamiliar word took down every conversation it appeared in.
A transcript is append-only and permanent, so the set of kinds one can hold
only ever grows: what this build writes is not what it may have to read. A
line can come from a newer server, or from an older one that wrote a kind
since dropped, and neither may be able to end the file.
`Indexed::parse_at` degrades a line it cannot make sense of to
`Event::Unreadable { kind }` instead of failing the whole read. It keeps the
line's seq -- the cursors, the page bisection and the next-seq counter are all
addressed by it, and dropping the line would hand out a seq the file already
contains -- and carries the word the line called itself, so the phone can say
what is missing rather than that something is. A line with no readable seq is
still an error: that one cannot be placed at all.
`Event::TaskNote` comes back retired rather than deleted: deserializable,
never constructed, dated, with the reason on it. The phone folds it to no row,
which is the point -- an unreadable line correctly draws a placeholder, and
one per background task is the wall the row was removed for in the first
place.
Found while diagnosing a report that live sessions had lost their status and
could not be sent to. 173 server tests pass, including the new one, which
fails on the old code within a second.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A row per finished background task is a screenful of dividers about work the
reader was not asking after, and one of them turned out to be a whole shell
command drawn as centred prose, because its words came from somewhere with no
reason to keep them short. `Event::TaskNote` is gone entirely, along with the
row that drew it. A subagent's closing report is recorded as that subagent's
own transcript's closing text and is read in the subcard, which is where it
was already going; what the parent gets a row for is a message a subagent
genuinely sends it, which arrives by the peer path and has had one all along.
What remains is the actual defect and the smallest thing that fixes it. The
fold still refuses to grow a settled reply, so a turn boundary is always a
message boundary, and where two replies then abut it puts a `TurnBreak`
between them: a hairline, no words, no colour. Made by the fold rather than
sent by the server, because it is not something that happened -- it is the
boundary between two things that did. `joinPages` puts one in at a page seam,
which the fold never gets to see.
The task notification is still what closes a task in `Status::Waiting`'s
bookkeeping, and the registry lookup that recognises one this translator never
saw start is what makes that work for a session adopted across a restart.
Verified on the emulator: three replies, three rules, and nothing about the
helpers anywhere in the parent. 170 server tests, 85 JVM tests, ktfmt, clippy,
rustfmt and Android lint clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A turn started by something with no row of its own -- a subagent reporting
back, a peer message the CLI only owns up to at the end -- met the previous
reply with nothing between it, and the fold grew that reply rather than
starting a new one. Two answers were drawn as one paragraph, running together
mid-sentence with not even a space between them. The fold now refuses to grow
a settled reply, and `joinPages` carries the same rule across a page boundary.
The other half is the row. `Event::TaskNote` records a background task
reporting back -- a subagent that finished, or a backgrounded command -- with
its title, how it ended and what it said; `TaskNoteRow` draws it as a card,
since somebody said this, and its own row rather than an update to the Task
call's, which is above everything the session has said since. Reported once
however many of the CLI's two lifecycle shapes arrive.
`SessionStatus::Waiting` is a session whose own turn is over while work it
started is not. `Idle` means "waiting for a person" and this means the
opposite, so reporting it as idle sent a "finished" notification at the one
moment that was untrue. Drawn as "waiting" in `waitingColor`; the queue and
the held-command boundary release on either end-of-turn status, so a message
sent while a subagent runs is not held until it finishes.
And a usage limit the account hits inside a subagent now reaches the session
as well as the subagent's transcript. `resume.rs` can only schedule against a
session, and a background Task outliving its parent's turn is the ordinary
case, so auto-resume was doing nothing at all for it.
The status word and its colour were two `when`s on two screens, and the second
missed `waiting` silently; they are `sessionStatusWord`/`sessionStatusColour`
now. Echo's `/subagent n` reproduces the whole shape, staggered a second
apart. Verified on the emulator against the sandbox: 169 server tests, ktfmt,
clippy, rustfmt, Android lint and the JVM unit tests all clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Off by default and per session: it spends quota the moment quota exists,
with nobody watching, which is not a thing a default may decide. Switched
on from the session settings dialog, with the message it sends editable
("continue" unless something else is typed).
Running out of quota becomes a state rather than an error. The Claude
driver recognises its dialect's sentence -- `Claude AI usage limit
reached|1788546972` -- and reports `LimitReached` with the reset time it
gave; nothing above a driver matches on a string. The transcript draws it
as a divider, like a clear or a compaction.
The schedule is a plan to *ask*, never a plan to send. Both reset times
available are untrustworthy in the direction that matters -- the dialect's
is written when the turn fails, the endpoint's moves when the window does
-- so the wait ends in a question to the usage meter, and only `ok` with
no window at 100% sends anything. A window still spent reschedules to its
own reset time, which is what makes a limit that lifts late wait longer
and one that lifts early resume sooner. A meter that cannot be asked is a
longer wait too, never a send. A day after the limit was hit the wait
gives up and says so in the transcript, so a machine that can never be
asked is not retried for ever.
The schedule is persisted on the session: a five-hour window outlasts a
backend restart, and a wait forgotten across one never comes back.
Driven end to end with echo, never a real account: `/limit [minutes]`
reports the same event a real driver does and `/usage` sets what the meter
answers, deliberately separate so the two can disagree. The wait moved
from the dialect's two minutes to the meter's seven when the meter changed
its mind, and the message went out on the first check after the meter came
back under the limit.
Also makes the settings dialog scrollable, which these two controls made
necessary: at a 1.5x system font it clipped the last of them with nothing
on screen to say so.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The same pass the server had, on the Kotlin side: comments restating what
the code says are gone, and the ones recording a measurement, a constraint
or an incident are kept but cut to a few lines each. 6540 comment lines to
5674, and 920 lines off the app.
Two doc comments had drifted onto the item above the one they describe --
`contextAfter`'s onto `sessionWorking` in Events.kt, and `UsageMonitor`'s
equivalent on the server was fixed in the previous commit. Each is back on
its own item, which is the only non-comment line this diff moves.
The comments are reflowed to the column limit at their own indentation:
several were written wide, and ktfmt re-wrapped them into lines holding a
single orphan word. `/tmp` script, not kept -- ktfmt is idempotent over the
result, which is the check.
Left alone deliberately: this codebase's remaining comment density is high
because the comments carry things the code cannot say -- what a null means,
what a number was measured against, which bug a guard exists for. Of the
238 one-line doc comments in the app, five were pure restatement of the
name and were removed; the rest each say something the signature does not.
ktfmtFormat, compileDebugKotlin, lintDebug and testDebugUnitTest pass;
cargo test (127), clippy --all-targets and fmt still clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Attachments were images only. Now any file can be attached: from the file
chooser behind the "+" menu, or from Android's share sheet, which the app
is now in. An image still goes to the model as a picture; anything else is
stored under its own name (`<hex>-<name>`, cleaned by `safe_file_name`)
and the Claude driver ends the message with `Attached file: /abs/path`,
since the CLI reads files by path and a model cannot be shown a trace. The
user-message field is renamed `images` -> `attachments` on both sides,
with a serde alias reading the rows written before. A share arrives before
anyone has said which session it is for, so it is held in AppRoot with a
banner on the list until a session takes it; an open session takes it at
once. Unreadable shares are reported beside the composer, not thrown.
The tool card crashed the app when opened on a command holding a quoted
glob such as `-path '*/.git/*'`: highlights 1.1.0's shell lexer answers
`x '*/a/*'` with a span whose end is before its start, and AnnotatedString
refuses the range. Such spans are dropped; the library is the place for
the fix. The echo driver gains `/bash <command>` so a card with a given
command can be produced on the emulator.
ui-sandbox.sh's token salvage read the tokens block's close only at a line
start, ran past the compact `),],` the server writes, and copied `setups`
into the new config twice, which the server then refused.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The transcript's remaining lag was the newest assistant reply: transcriptUnits
kept the last row whole -- right while it streams (splitting a changing text
is a parse per delta), wrong forever after, so a session that ends on a long
reply drew it as one lazy-list item with every node alive. On a Pixel 9 Pro
XL that was 13.8ms of draw phase a frame, 79% of it the framework's own
per-node bookkeeping, against a 34,996px item.
An AssistantMsg now carries `settled`, folded from the status event that ends
its turn (status changes are transcript events with seqs, so replay settles
the same way), and cleared if a delta ever grows the message again. A settled
newest reply splits like every other. Folding it -- rather than reading the
screen's status -- routes the resplit through the held-events gate, so it can
only happen at the newest end while pinned, never under a reader. The
"session is working" predicate now lives once, in sessionWorking().
Measured on the emulator, same session and gestures, a 43KB reply as the
last row: draw phase 3.92ms -> 1.20ms per frame, framework share 3.07ms
(78%) -> 0.54ms (45%), worst single measure 82.5ms -> 9.1ms. The report's
"on screen" line went from one 60,674px AssistantMsg to five blocks of
95-846px. A live streamed turn settles and splits the moment it goes idle.
The harness half, asked for by Bryan: ui-sandbox.sh now derives its port and
root from the checkout name (two checkouts' sandboxes cannot reach each
other), keeps its token in ~/.config/ai-app/sandbox-token and salvages
enrolled device tokens across restarts (enrol the emulator once, ever), and
gained the driving verbs every UI session was re-inventing in /tmp: spawn,
send (text or @file), api. transcript-bench.sh is the standard
scroll-and-report measurement. AGENTS.md documents all of it.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The live Claude Code path only learns a turn was another agent's when
the turn ends -- the whole of the message arrives as an `origin` object
on the `result` -- so the note was appended after everything it caused,
and the transcript showed the answer above the question.
It cannot be recorded in place: by the time anyone knows, the reply is
already written, and the transcript is append-only. So the event carries
where it belongs instead. `PeerMessage` gains `turnStart`, the seq of
the status that opened its turn, stamped by the pump -- the only thing
that knows a seq and the only thing that sees every driver's turns. The
phone gives the note that seq, so it sorts into place rather than being
drawn out of order at the end. A status draws no row, so there is
nothing for it to collide with and the list stays sorted, which the
scroll anchor and paging both depend on.
Absent where there is nothing to correct: a message replayed out of a
session file by `import` is already in the right place, and one that
opened no turn has no turn to sit above. Both stay where they arrive.
The echo driver gets `/peer-turn` for the live shape, beside `/peer` for
the in-place one. Verified on the emulator both ways, live and on
replay, plus an ordinary `/tools` turn to confirm the run grouping the
insertion cuts through is unaffected.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Two things a reader could not do to what is on screen.
**Selection.** Nothing in the transcript was selectable at all, so a
command, a path or an error message could be read and not copied. One
`SelectionContainer` around the whole list rather than one per row: a
transcript is one body of text to a reader, and a selection has to be able
to run from a reply into the tool output under it. Per row it also could
not, and whatever was drawn without a container would have been silently
unselectable -- a state nothing on screen reports. Rows keep their tap
handlers; checked on the emulator that expanding a tool call, scrolling and
flinging are all unaffected, since a selection is a long press.
**Taking a message back.** A message sent into a running turn sits as a
bubble waiting to be read, and there was no way to change your mind: it is
tappable now, and the server answers `POST /sessions/{id}/unqueue`.
The answer has three states, and the middle one is the point. Claude's
driver writes a steer into the CLI's stdin the instant it arrives -- that
is what makes it reach the model at the next tool boundary rather than at
the end of the turn, and it was measured -- so the line is already gone and
`AlreadySent` is the only honest answer it can give. Holding the write
until a boundary would make the drop real and cost a steer one model call,
which is the latency the immediate write exists to remove; rejected on that
trade, with the reasoning in PLAN.md. The refusal is drawn on the bubble
that was pressed rather than in the error row under the header, a screen
away from it.
Where a driver really does hold its queue -- echo today -- the message goes
for good, and it goes as an `Event::MessageDropped` rather than as a return
value: every device watching the session loses the bubble, and a phone that
reconnects and replays the `messageQueued` does not put back one that was
cancelled with nothing left to resolve it.
A Rust backend that owns the sessions and an Android app that reads them.
The server spawns and adopts CLI processes, normalises everything they emit
into one event model, keeps the transcript, and serves it over pinned TLS on
a WireGuard interface; the phone streams that, replies, sends images, and
imports conversations the machine already has.
`AGENTS.md` is the working guide -- what runs where, what has been measured,
and the faults that were expensive to find. `PLAN.md` is the design record.
History before this point was squashed away. It was a personal project's
running commentary and carried a name and a couple of machine paths that
have no business in a public repository; the tree is what mattered and the
tree is here.
iris-aiandClaude Opus 5
iris