A new Claude session names its model but not its effort level, so the
hook fell back to the settings' level - and a session started on Haiku
showed "High" until its first turn ended. The hook now checks Claude
Code's own model catalog (its cache, read only) first: a model it lists
without effort options is reported with an empty level. A model it does
not list, or no catalog at all, falls back to the settings as before.
While a hook runs Gemini CLI shows "Executing Hook: <name>" above its
input, and a hook without a name is shown by its whole command - the
quoted path of the tty7 binary and its arguments. tty7's Gemini entries
now carry "name": "tty7".
An install from before reads as outdated - its hooks' names are checked
along with their commands - so the launch-time refresh and the settings
page's Update rewrite it in place, one entry per event as before. Other
agents' entries get no name and are judged as before.
A file picked from the composer's @ menu went in raw for Claude Code and
Codex, so "my notes.md" came out as "@my notes.md":
- Claude Code ends a bare mention at the first whitespace, so it looked
for "my" and the file never reached the turn. A path with a space now
goes in double quotes, @"my notes.md", the form Claude's own list puts
in and its parser reads back as one file.
- Codex hands mentions to the model as plain text, and its own list puts
in the path alone: without the @, in double quotes when it has a space
in it. The composer now does the same.
Checked for real: Claude 2.1.286 with no tools answers from @"my notes.md"
and not from @my notes.md or @my\ notes.md; Codex 0.158's list inserts
"my notes.md". Attachments are unchanged; both agents read one with a
space in its path.
Gemini CLI judges each key of a typed burst against its input as it was
before the burst, so to it the input is still empty at every key. A "?"
anywhere in a typed line is taken for the key that opens its shortcuts
on an empty input and dropped ("what? yes" arrived as "what yes"), and a
"!" for the one that switches it into shell mode, which also clears what
came before it: "hi! there" became a shell command " there", one Enter
away from running.
Gemini now gets every message as a bracketed paste, like Codex, with a
leading "!" still sent as a key of its own - as for Claude Code - so a
message that starts with one keeps opening shell mode.
Gemini CLI now names its session browser /resume, with /chat kept as an
alias, and /clear starts a new session rather than only clearing the
screen. Offer /resume under its own name, say what /clear does now, and
add /model and /init, which the menu left out.
Three things kept files from reaching a Gemini turn from the composer:
- Attachments went after the text as shell words. Gemini turns a drop
into @ mentions itself, but only a paste that is nothing but paths;
after the message's text a path is just words, and the model never saw
the file. Gemini now gets its attachments as the mentions it would have
made.
- A mention picked from the @ menu was spelled raw. Gemini ends a mention
at the first unescaped space, so "my notes.txt" became "@my". Mentions
for Gemini are escaped the way its own escapePath does it.
- With the caret at the end of an @ word Gemini lists matching files, and
Enter takes the list's pick instead of sending, so a message closed by
a mention - every message with an attachment - sat in the input under
the box, unsent. A message with a mention now ends in a space, which
closes the list; slash commands are left alone.
Gemini CLI takes an Enter that comes within 40ms of a bracketed paste as
a line break. It starts that window once it has read the whole paste, so
the 50ms the composer left between the paste and the Enter was not
enough for a long message: a 41-line message sat in Gemini's input as
"[Pasted Text: 41 lines]" plus an empty line, hidden under the box, and
was never sent. Gemini now waits as long after a paste as Copilot does;
a 400-line, 32KB message goes through whole and is submitted once.
Gemini CLI no longer frames its input in a rounded box: it shades the
prompt line between a row of lower half blocks and a row of upper half
blocks, and with shading off it draws a rule over the prompt instead. The
composer looked only for the old frame, so over a current Gemini it
stayed stepped aside for good and never covered the input.
A message already sent is shaded the same way in the transcript, so the
block counts only with nothing but the footer under it; a permission
prompt or a list in the input's place leaves the last message above more
than that, and the box steps aside for it. The old frame is still found
for older versions, and the shell, YOLO and history-search marks count as
a prompt as well as `>`.
Gemini CLI roots its user-level .gemini directory at GEMINI_CLI_HOME in
place of the home directory. tty7 always wrote and checked
~/.gemini/settings.json, so for such a user the hooks it installed never
ran, and the state it reported described a file Gemini does not read.
Local-only, like the other config-dir overrides.
Codex runs its hooks in its app-server, one background process every
Codex session on the machine talks to. The first session starts it, and
it outlives that session, detached. Its hooks therefore carry the first
session's $TTY7_PANE and run under that pane's shell only while that
session lasts. Once it ended, the daemon refused every Codex report as
coming from outside the pane -- and the tty fallback has no terminal in
a detached process -- so a Codex pane never showed working, waiting or
done. While the first session was still running, a second pane's
reports were taken as the first pane's.
A Codex report is now matched to its pane by what it says: the pane
that already reported its session, else the one pane running Codex in
its directory (the one yet to report a session, and the named pane on a
tie). Other agents' reports are unchanged.
Codex reads hooks.json from its home, which CODEX_HOME moves. tty7
always wrote and checked ~/.codex/hooks.json, so for such a user the
hooks it installed never ran, and the state it reported described a file
Codex does not read. Local-only, like the other config-dir overrides.
Codex marks the selected row of its lists -- the approval prompt,
/model, /permissions, the hook review and the update offer at launch --
with the same `›` that starts its input line. The composer took that row
for the input and laid the box over the list, hiding the options the
user had to pick from. A `›` row that is a numbered option with sibling
options around it is now read as a list, and the box steps aside.
The toolbar's effort picker types `/effort`, but the `/` menu did not
list it. Bring the list in line with current Claude Code while at it:
`/usage` replaces `/cost`, which is now only its alias, `/rewind` joins,
and `/agents`, which Claude Code has removed, goes.
An agent hook reported by writing an OSC 777 sequence to the tty the
agent draws on. A tty write is not atomic: on macOS the kernel parks a
writer whenever the output queue is over its high-water mark, which is
exactly when the agent is busy redrawing, and lets the other writer in.
The agent's output then lands inside our sequence, the OSC breaks, and
the rest of the JSON is printed on screen. A pty test with a busy
writer next to a small one splits even 100-byte writes, at arbitrary
offsets, so no size keeps the sequence whole.
The hook now sends the same JSON to the daemon that owns its pane
(`$TTY7_PANE`, over the pane socket `$TTY7_CONFIG_DIR` leads to), which
applies it exactly as if its reader had found it in the output. The
pane id comes from the environment, which a detached tmux server or
the like carries out of the pane, so the daemon applies a report only
from a process that runs under the pane's shell. Anything else — no
daemon, one that predates the message, a refusal — falls back to the
tty as before, which is still how a hook with no daemon beside it
reports.
A hook reports to the pane by writing an OSC 777 sequence to the tty the
agent draws on. macOS does not keep a tty write whole: under output
pressure the kernel parks the writer mid-sequence and lets the other
writer in, so the agent's redraw lands inside our sequence, the OSC
breaks, and the rest of the JSON is printed on screen. The longer the
sequence, the likelier that is.
The prompt a turn starts with was the bulk of it on `prompt-submit` (up
to 200 characters, three bytes each for CJK), and nothing reads it any
more since the conversation outline went away. Stop sending it, and cap
the agent's message at a status line's worth.
A model or effort picked from the toolbar was shown until the agent
reported a different value. When the pick did not take — Claude asks
before switching a cached conversation's model, and "No, go back" keeps
the old one — the report never changed and the toolbar showed the refused
model for good; after the agent quit and a new session started on the
model the pick was made from, the stale pick came back.
A pick now stands until the next finished turn, which reports what the
agent actually ran on, and is dropped when the agent session changes.
Claude Code's `/model` list has Fable between Opus and Sonnet, and
`/model fable` takes it by that alias, but the toolbar's list left it out:
it could not be picked there, and a session running it had no row checked.
The hook filled in the model from the transcript's last reply on every
event. That reply is the previous turn's: after a `/model`, or a resume
under another `--model` (a resumed session's SessionStart names none), the
toolbar showed the old model — and at `Stop` too, since Claude Code runs
the hook before the turn's reply is always on disk. The model now comes
from SessionStart, or at Stop from the reply that turn wrote: found after
the prompt the payload names, matching the words it carries, waited for
briefly. Nothing exact means nothing reported.
Effort likewise: the settings fallback only runs for a model the event
names (it may be set per model), and a turn that ends without a level ran
on a model that takes none, which the toolbar now shows as no level rather
than the settings' general one. A level picked from the toolbar gives way
to the agent's report once that changes, as a picked model does.
The covering layer measured its height up from the pane's bottom edge in
whole rows plus the padding, but the grid's rows start at the top: whatever
part of a row the pane's height leaves over sits between the last row and
the padding. At most heights that left the top of the covered area — Claude
Code's upper rule — showing above the box. The grid now records that
leftover and the layer counts it in.
With the box laid over the agent's input, typing at the grid already went
into the box, but a paste (Cmd+V, a dropped or copied file) still went to
the pty — into the agent's own input, out of sight under the box, where the
next Enter would send it unseen. A paste now lands in the box the way
typing does, files as attachments, and takes the keyboard with it.
The toolbar's thinking toggle is gone; reasoning effort is what the
agent actually exposes as a per-session dial. The button shows the level
Claude Code is set to (CLAUDE_CODE_EFFORT_LEVEL, else effortLevel from its
settings, reported by the hook) and opens a menu of low / medium / high /
xhigh / max. Picking one sends the agent's own /effort command, and the
label follows the pick until the next hook event confirms it.
The composer now stands in for the agent's input instead of sitting under
it, so the pane has one input rather than two.
- For Claude Code, Codex and Gemini the box is laid over the input area it
finds on the grid (Claude's ruled prompt block and the status rows under it,
Codex's `›` line, Gemini's framed prompt), painted in the grid's background.
When the agent puts something else there — a permission prompt, a picker —
the area stops reading as an input; after a 250ms grace the box steps aside
and the keyboard goes to the TUI, and both come back with the input. Other
agents keep the docked layout.
- Over the input, Esc is the agent's (it interrupts), Ctrl+C clears the box or
reaches the agent when the box is empty, and text typed at the grid lands in
the box.
- The toolbar: permission mode (read off the status row the box covers, falling
back to the hooks' permission_mode; click or Shift+Tab to cycle), model
(click opens Claude's picker), Think (Claude's thinking toggle), and a context
ring. Labels come only from what the agent reported.
- Claude hooks now carry an AgentReadout — permission_mode from the hook
payload, the model and context size of the last main-thread reply from the
transcript, the context window (1M for a [1m] model or past 200k), and
alwaysThinkingEnabled — worked out in the hook so remote panes get it too.
- Visual pass against the design: 13.5px text with the design's padding
(the input's own padding subtracted), a 40px toolbar of 28px buttons, a dark
round send button, image chips with their own glyph.
- A rounded, tinted frame whose hairline darkens on focus and on a file drag,
a toolbar row with an attach button and a round send button that fills in
once there is something to send. The key hints move into the send button's
tooltip; the placeholder names the agent and what / and @ do.
- Files become chips instead of words in the text: the attach button, a drop
on the box, a pasted screenshot or copied file, and remote uploads all go
through one paste_paths route that attaches while the box has focus.
Backspace at the start of an empty caret removes the last chip. On send the
paths follow the text as the shell words a drop would have typed.
- `/` at the start of the message opens the agent's built-in commands (Claude
Code, Codex, Gemini) plus Claude's custom commands from .claude/commands;
`@` at the start of any word opens files from the pane's project, reusing
the quick-open walk so it works on remote hosts too, spelled relative to the
pane's directory. Up/Down move, Enter/Tab pick, Esc dismisses.
- Shift+Tab is passed through to the agent to cycle its permission mode.
- Enter with nothing to send no longer sends a bare Enter.
No model picker, mode label or context meter: tty7 has no source for them
that would stay true, and /model is one keystroke away in the menu.
A multi-line text box docked at the bottom of a pane whose foreground is a
coding agent, for writing a prompt with the platform's own editing keys, mouse
selection and an in-place IME, then handing it over whole. Toggle with
Cmd+I (Ctrl+Shift+I off macOS) or "Toggle Message Composer" in the palette.
- Docks below the grid instead of floating over it, so the agent reflows into
the remaining rows and nothing it draws is hidden.
- Enter sends, Shift+Enter inserts a newline, Esc hands focus back to the
terminal without closing (so the next Esc still interrupts the agent).
- A message is written as text, then Enter as a separate write after a short
settle. Multi-line text goes as one bracketed paste; Codex is always pasted
(its burst detector swallows the Enter otherwise); a leading `!` reaches
Claude Code as its own key so it switches to shell mode.
- Nothing is sent while the agent is waiting on a permission prompt or
question; the box turns amber and keeps the message.
- Writes queue per pane and stop if the agent exits mid-send, so an Enter can
never land in the shell.
- Pasted screenshots, copied files and dropped files reuse the terminal's
existing path-pasting routes and land in the box when it has focus.
- Drafts and the open state survive the view being rebuilt.
Refs #842
* feat(panel): CPU% and memory per process, and a total, in Info → Processes
The daemon adds rss, cpu_ns and a start stamp to each ProcEntry (serde
default, so an old daemon or GUI skips them). The GUI turns two CPU-time
samples into a ps-style % and sums both into a Total line. `tty7 procs`
gains an RSS column.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(panel): no CPU% across a poll gap, and test CPU time is in ns
A new poll chain starts after the panel was shut or the pane changed;
comparing against the sample from before the gap showed a long-run average
as the current figure. Reset the tracker so the first round shows a dash.
Replace the self-usage test's cpu_ns > 0 (fails on Linux before the first
10 ms tick, passes unconverted mach ticks) with a check that the process's
CPU time covers this thread's own burned CPU time.
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
* fix(sidebar): Move to Group is a submenu listing every sidebar group
The tab menu's Move to Group listed only pinned groups, laid out flat on
the grounds that there are never many. A sidebar that auto-groups by repo
holds dozens of groups, none of which could be picked. It is now a submenu
with every group the sidebar draws, in sidebar order, the tab's current one
checked, then Ungrouped (back to auto grouping) and New Group. Picking an
auto group pins it, as its header's pin does, and keeps the tab there.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(sidebar): Move to Group review fixes
- Ungrouped row replaced by Remove from Group, enabled only for a tab kept
in a group; move_targets is pure (no locale lookup) over GroupKey.
- move_tab_to: an auto group is pinned with the tab in one edit
(pin_auto_group_with), gpui-tested.
- A separator splits pinned from auto groups; targets are taken before the
submenu borrows cx.
- No CHANGELOG hunk; the description carries it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(sidebar): Move to Group comment and pin_auto_group_with tidy
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(sidebar): Move to Group's labels line up with the tab menu
One left-checked row makes gpui_component reserve a check column on every
row of that menu, so the whole submenu sat an icon's width right of the tab
menu it hangs off. Putting the check on the right keeps the labels at the
parent's inset.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(sidebar): pinning a repo whose folder is kept joins the kept group
A tab dragged out of a folder group while still in the repo is drawn
under the repo's auto group, beside the folder group keeping the same
directory. Move to Group lists both; picking the auto one (or its
header's pin) pushed a second folder group on the same path, two
identical headers splitting the folder's tabs by list order. Join the
kept group instead, as pin_folder already does.
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
* feat(terminal): program notifications follow the policy; read kitty OSC 99
OSC 9/777 notifications used to post from the reader thread regardless of
focus or the Never setting, and clicking one revealed nothing. They now queue
for the view, which applies notify_on_command_finish (Unfocused holds one back
only while its pane is focused in the key window), posts it clickable for the
pane, and titles it with the agent's name when it brings none. A pane whose
agent reports through tty7's hooks skips the copies the hooks already cover.
Kitty's OSC 99 is read too, chunked by i= with d=/p=/e=, in the client and in
the daemon sniffer, so a hookless agent's kitty notification marks it Waiting.
Claude Code's ghostty, kitty and iterm2 Notifications channels all land here;
its default, auto, sends nothing under TERM_PROGRAM=tty7.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(terminal): agent hook notices follow the per-pane rule
An agent's Waiting and Done notices from tty7's hooks were only posted while
the window was in the background, so an agent in another tab of the front
window never reached you. They now share shows_notification with program
notifications: Never posts nothing, When unfocused holds a notice back only
while its pane is focused in the key window, Always always posts. A hooked
pane still skips its program's own copies, so nothing doubles.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Revert "feat(terminal): agent hook notices follow the per-pane rule"
This reverts commit 7df400043c. The hook
notice change is a behaviour change of its own and moves to the stacked
branch upstream/hook-notices-per-pane.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor(terminal): NotifyMode::allows, a Note type, and a cap on queued notes
One rule for every notification path: a mode allows a notice unless it is
Unfocused and the reader is watching. Kitty's pending chunks and a
finished note get names instead of tuples, and a pane whose view is not
polling keeps only its newest eight notes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(changelog): drop the Unreleased entry; the release notes carry it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(terminal): a burst of program notes shows its newest three, and an exiting shell's are shown
`take_osc_notes` hands over the newest three queued notes and drops the
rest, so a program that writes a burst doesn't spray the desktop. The view
shows them ahead of `poll_foreground`'s exit check, so what a program said
just before its shell exited still gets through.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor(terminal): one cap on queued program notes, three
The queue kept 8 and `take_osc_notes` then handed over the newest 3: two
caps for one rule. The queue now keeps 3 and `take_osc_notes` drains it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(terminal): at most five program notes per pane every ten seconds
The queue cap applies between polls, so a pane that keeps writing
notifications still posted a few every 300ms. Each pane now has a
NoteBudget: at most five notes per ten seconds reach the desktop, and the
rest are dropped and said once, as "More notifications from this pane
weren't shown", when the window turns over. The budget is asked on every
poll, held-back ones included, so that note comes even after the flood
stops and never for a count gone stale.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(terminal): kitty control payloads and id-less chunks don't build a note; OSC 9;10-12 aren't notes
A p=close, p=alive or p=? with the id of an unfinished OSC 99 note
finished and posted it. They are commands about notifications, so they
now leave pending notes alone. Chunks without an i= are each their own
notification per the spec, and no longer join an earlier id-less d=0
chunk. ConEmu's OSC 9;10, 9;11 and 9;12 (a prompt mark some shells emit
every prompt) were posted as notifications; they are subcommands like
9;1-9;9.
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
* fix(terminal): tell a pane's program when the theme turns light or dark (DEC 2031)
Claude Code's `theme: auto` reads OSC 11 at startup and re-reads it only
when the terminal sends `CSI ? 997 ; 1|2 n`, which it asks for with
`CSI ? 2031 h`. tty7 ignored 2031, so a running Claude kept its light diff
colours after a flip to dark. Track 2031 per pane (and across reattach),
push a 997 when the theme's background changes, and answer `CSI ? 996 n`.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(terminal): report the colour scheme on reattach; review cleanups for DEC 2031
A pane reattached with 2031 on now hears the current scheme once the
replay is folded, so a theme flip while it was detached still reaches
the program. report() uses the presets' is_dark, watch reads
view.terminal directly, TRACKED lists 2031 inline, and the gpui test
moves into view.rs's gpui_tests harness with a reattach case.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(changelog): drop the Unreleased entry; the release notes carry it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(terminal): one scheme report per reattach, not one per replayed frame
A replay is a modes Snapshot plus one Snapshot per ring segment, and each
sent its own 997. The Snapshot arm now only folds and flags; the report
goes out once, on the first frame after the replay. The tests use the
harness's next_input_until_timeout instead of a reader thread, and the
reattach test replays two Snapshots.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(terminal): send the reattach scheme report when the replay has been read
It went out on the first frame after the replay, so an idle pane, with no
live frame coming, never heard it. It is now sent once the reader's buffer
drains. TRACKED keeps its one line and names `COLOR_SCHEME_UPDATES` rather
than repeating 2031. The reattach test replays Snapshot, Size, Snapshot,
Snapshot with nothing after.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(terminal): a shell prompt ends DEC 2031, so a dead program's mode never reports into the shell
A program that switched 2031 on and died without `?2031l` left the mode on
in the pane's tracked modes, its ring, and the client's fold. A reattach
replay then still ended with 2031 on, so the reader sent its post-replay
report, and every theme flip sent another, into the shell's command line.
`TerminalModes` now drops 2031 on an OSC 133 `A`, `B` or `D` mark: the
shell owns the terminal again, so whatever asked for reports is gone.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(terminal): send the reattach scheme report after the replay, not at a read boundary
The reattach report went out whenever the reader drained what it had read.
An 8 MiB ring arrives over many reads, so a read that ended after the ring
segment holding a dead program's `?2031h` and before the one holding the
shell prompt that ended it typed `CSI ? 997 ; n` into the shell. The report
now waits for the first frame past the ring (the replayed Prompt, Cwd and
the rest, or live Output once it is folded).
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
* fix(themes): accept integer colour components in .itermcolors files
iTerm2 and plistlib write exact 0 and 1 as <integer>, which the loader
rejected, dropping the whole theme.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(changelog): drop the Unreleased entry; the release notes carry it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(ui): selecting a sidebar tab no longer scrolls the list
activate() called sidebar_scroll.scroll_to_item(tab_index), but the
sidebar list's children are group blocks and dividers, not tab rows, so
the index named some other group and a click on a lower row jumped the
list. The row now brings itself into view when it is drawn: a row with
any part on screen stays put, one out of view scrolls in to the nearest
edge.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(ui): key the sidebar reveal by tab id and drop it when the row is not drawn
An index went stale when tabs were reordered or closed before the next
frame, and a reveal aimed at a row in a folded group fired whenever the
group was opened, long after the tab was selected.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(changelog): drop the Unreleased entry; the release notes carry it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(ui): a first row revealed from above brings its group header
When the sidebar scrolled a row in from above, the row landed on the top
edge with its group's heading still cut off above it, so the tab came into
view without the name of the group it is in. The first row of a group that
draws a header now counts the header (and the gap under it) as part of
itself when coming in from above. Rows on screen and rows below are
unchanged.
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: l0ng-ai <24760907+l0ng-ai@users.noreply.github.com>
An agent reporting a file it wrote prints `原型:/tmp/a.html` or
`原型：/tmp/a.html` with no space. The label rule only accepted ASCII
labels behind an ASCII colon, so the whole token was read as a relative
path and the click reported it missing under the pane's directory.
Accept any alphabetic label, one character long when it is not ASCII,
behind either colon. A single ASCII letter stays a Windows drive.
Closing the last file left nothing in the editor panel to hold the focus,
so it fell back to the terminal pane and the next ⌘W closed the terminal
instead of the empty editor.
The empty panel now has its own focus handle: it takes the focus over in
the frame after its last file closes (only for the tab whose editor had
it), and a click on it focuses it. ⌘W from there, or from the file tree
an empty ⌘⇧E hands the focus to, hides the panel and returns the focus to
the terminal. With the focus in the terminal, ⌘W is still the terminal's.
Claude-Session: https://claude.ai/code/session_011mDkkQhwx4RJJBHee3yVpq
* fix(mobile): bump iroh to 1.3.0 so a stuck relay cannot stall direct dials
iroh 1.2.0 sends a connection's first datagrams to every known path one
after another inside the remote-state actor, awaiting each. When the relay
is unreachable its send queue fills and the actor blocks, so handshake
packets for a direct path that does work (a mesh VPN address, a public
IPv6) queue behind it and the dial times out. 1.3.0 sends to all paths
concurrently with a bounded wait (n0-computer/iroh#4512).
The desktop workspace was already on 1.3.0; the app has its own lockfile
and was left behind.
* fix(gateway): reach the relay through a proxy when that is the only way out
iroh dials its relay with its own resolver and its own TCP, ignoring the
system proxy. On a machine whose network only works through a local proxy
(Clash and the like, system-proxy or TUN mode alike) that dial never
succeeds, and nothing says so: phones on the same network still connect,
while a phone on cellular times out, because without a relay nothing
coordinates hole punching and the home router drops unsolicited inbound
packets.
The gateway now lists the ways out it knows of, most likely first:
tty7's own http_proxy setting, the system proxy, the environment's, then
direct. It starts on the first without waiting, and when the relay stays
unreachable for 10s it tries the others on a throwaway endpoint, switching
to the first that reaches a relay (same key, same port, so pairing codes
keep working). While none does, it looks again every minute.
A pairing code now names the relay only once it is actually connected;
before, it named whichever relay iroh picked by latency, reachable or not.
The platform proxy readers in daemon::install::proxy now also hand back
the proxy as a URL, for a client that is not ureq.
Claude-Session: https://claude.ai/code/session_018wE9ZRyxgWW2f55VSy9FvZ
Built from the files the night actually produced, so one that did not build is left out rather than linked dead. Links go through the rolling nightly tag.
Release and nightly now build the desktop app for aarch64-pc-windows-msvc
alongside x64, publishing tty7-<version>-windows-arm64-setup.exe and
tty7-<version>-windows-arm64.zip.
- Cross-compiled on the x64 Windows runner, like server-windows' ARM64
leg, and marked experimental until it has shipped once: a failure drops
the ARM64 packages instead of holding up the release.
- Vendors Microsoft's arm64 ConPTY pair from the same package version as
the x64 one, so ARM64 panes keep the OSC 11 fix (#345).
- Each Windows package bundles the WSL server for its own architecture.
- The installer's architecture comes in as a define: x64compatible for the
x64 build (still installable on ARM64 under emulation), arm64 for the
native one. One AppId, so either upgrades the other.
- The in-app updater on an ARM64 build asks for the arm64 packages; an x64
build under emulation keeps taking x64 ones.
Claude-Session: https://claude.ai/code/session_011mDkkQhwx4RJJBHee3yVpq
l0ng-ai
