Review fixes for the charging-case battery:
- The case reports whether the glasses are docked. An empty caseInfo
(the reply out of the case) now parses as "not docked, no level"
instead of nothing, and undocking is reported even when the level and
charge state are unchanged. TS keeps the last level but clears
caseCharging while the glasses are out of the case.
- The phone line shows the charging mark only while it is live: G2 and
Case only with a link to the glasses (connected or charging), and Case
only while docked. Levels stay as the last known ones.
- The session-start case read waits until the layout is up, so a
firmware that ignored it could no longer hold up the first frame. When
the glasses are already charging at connect, the charging-mode read is
the only one; the old duplicate read and its "late ACK" log are gone.
- Case dedup state resets on every connect, as wearState does, so each
session's read reaches TS.
- caseCharging uses > 0 like the headset, and is normalized to 0/1.
- iOS now feeds the reading into the shell battery state, and the
assistant's describeState includes the case level.
- The batteryLine accessors move above connectionStatusLabel's doc
comment.
A line under the mirror lists each device whose level is known, e.g.
"G2 100% · Case 98% ⚡ · R1 72%" (⚡ = charging), in portrait and
landscape. The case level joins the shell's battery state beside the
glasses and R1 (it is not drawn in the glasses' top bar) and reaches the
page through the dashboard snapshot. Like the glasses level it is kept
across disconnects: the case only reports while the glasses are docked
and only drains while charging them.
Waking into the app switcher stays the default. With a popup switcher whose
status bar shows only in the switcher, waking into the app would hide the
clock and notification icons on every wake.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A "Hermes agent" assistant model talks to a Hermes Agent gateway's
OpenAI-compatible API server (streamed /v1/chat/completions), with Hermes URL
and key settings under Assistant. Hermes keeps the conversation: each
assistant session maps to one X-Hermes-Session-Id and only the latest user
message is sent. Faceclaw's own tools are not offered to it, and its reasoning
stream is not shown on the glasses. See notes/hermes-assistant.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- "Whisper server (self-hosted)" transcription provider for OpenAI-compatible
Whisper servers (vLLM, speaches, LocalAI) over plain HTTP batch requests,
with URL, model, language and key settings (Voice). Pause-ended spans are
one request each, finalized in audio order, with provisional text while a
span is open. See notes/whisper-server.md.
- Transcribe (and other continuous capture) works on iOS with a cloud or
Whisper-server provider; push-to-talk shares a running capture.
- Transcribe saves each session's audio as transcribe-<time>.wav in Downloads
(setting "Save Transcribe audio", on by default), written as it is captured
so a crash leaves a playable file; a saved transcript uses the same stamp.
- Calculator stops continuous listening when its window closes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A double tap that wakes the display (a screen-off double-click, or the
firmware's display-wake) now focuses the foreground window instead of the app
switcher. "Wake into last app" (Display settings, on by default) turns it back
to the switcher. Lock screen, put-on, notification and wakeword wakes are
unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>