redefine targeting: 'set' (aliases sticky/switch) is the persistent sticky
default (~/.claude-active); 'target <name> <command>' is a one-shot override that
routes a single command without changing the sticky default. add 'unset' and
'list'. resolution moves to a single target.resolve(one_shot) implementing the
order: one-shot -> sticky-if-exists -> only-session auto -> ambiguous/none do
nothing (never falls through, never injects into a missing session).
grammar.parse now returns ParsedCommand(one_shot, action) and skips optional
leading filler words (config behavior.filler_words: select/use/choose), with a
filler-before-digit still meaning the select command. CLI gains set/unset/list
(switch kept as a set alias). daemon console shows the targeting reason per line.
docs updated; no stale 'target = sticky' wording remains. bump to 0.1.1.
Signed-off-by: disqualifier <dev@disqualifier.me>
terminal-run is the product, so remove all backgrounding: delete the
claudedo.service unit and autostart.sh, strip the systemd step and the
autostart source-line from install.sh (rc block now sources cc.sh only).
claudedo start now runs a mic check first (warm-up + brief capture, aborts with
guidance if silent; --skip-audio-check to bypass) then drops into a visible
listen loop printing the recognition/action log: a startup banner, then
heard -> matched -> target / injected per utterance, target/mode state changes,
and (listen mode) non-wake speech dropped WITHOUT the transcript per the privacy
invariant.
Signed-off-by: disqualifier <dev@disqualifier.me>
WSLg's RDPSource suspends when idle and emits ~1-2s of silence while it resumes
on the first read, so a short timed capture (test-audio) or the first utterance
after daemon start could be lost. add audio.warm_up() that opens a stream and
reads until a non-silent block arrives (or times out); call it at daemon startup
and before test-audio's capture. test-audio now primes then captures 3s.
Signed-off-by: disqualifier <dev@disqualifier.me>
extract an OutputHandler abstract base; TmuxOutputHandler is production
(send-keys, PTY-only), StdoutOutputHandler prints what would be injected so
grammar+keymap run end-to-end without a live claude session (the deterministic
test path). module-level shims default to tmux so the daemon is unchanged.
Signed-off-by: disqualifier <dev@disqualifier.me>
append escape-time 0, large history-limit, allow-passthrough, and extended-keys
to ~/.tmux.conf under an idempotent marker block (no clobber). required for
reliable keystroke injection and for notifications/modified-keys to reach the
claude pane.
Signed-off-by: disqualifier <dev@disqualifier.me>
state terminal-run as the product (the claudedo start terminal is the
recognition/action console) and frame backgrounding/autostart/systemd as
optional extras, not the default.
Signed-off-by: disqualifier <dev@disqualifier.me>
remove inline comments (CLAUDE.md: docstrings only), strip trailing periods
from single-line docstrings, and fix a PulseArmy->PulseAudio typo. no behavior
change.
Signed-off-by: disqualifier <dev@disqualifier.me>
cc kit as a sourced ~/.config/claudedo/cc.sh (bash+zsh, forced explicit names).
opt-in rc autostart guarded by CLAUDEDO_AUTOSTART + an optional systemd user
unit. install.sh is idempotent: WSL audio deps, ~/.asoundrc pulse shim, audio
verify, model prime, and source-line rc wiring with backups.
Signed-off-by: disqualifier <dev@disqualifier.me>
daemon.py runs the loop with pidfile/state, runtime mode switching, and the
privacy invariant: in listen mode any non-wake utterance is dropped the instant
grammar.parse() returns None. __main__.py exposes start|stop|status|test-audio|
install|switch.
Signed-off-by: disqualifier <dev@disqualifier.me>
word-boundary wake stripping that's lenient on the coined word 'claudedo'
(despaced-prefix match) without swallowing the command's spaces. data-driven
phrase->action map; number words normalized to digits; 'target' aliases
'switch'.
Signed-off-by: disqualifier <dev@disqualifier.me>
stt.py wraps faster-whisper for fully on-device transcription. audio.py
captures via sounddevice with two paths: silence-segmented for listen mode
and held-key for ptt. resolves the input device from config (auto/index/name).
Signed-off-by: disqualifier <dev@disqualifier.me>
keys.py is the single source of truth for claude code's prompt keystrokes,
confirmed empirically against a live v2.1.191 session: bare digits select-
and-confirm immediately (no trailing enter). inject.py only ever calls
tmux send-keys (PTY injection, never OS input). target.py resolves the
active session from ~/.claude-active with a has-session guard and a single
session_name() mapping (claude-<name>) shared with the cc kit.
Signed-off-by: disqualifier <dev@disqualifier.me>
load/validate config.toml with clear errors; defaults to listen mode and
the 'small' whisper model. all tunables (wake phrases, audio thresholds,
type_autosend) live here, no hardcoded paths or secrets in code.
Signed-off-by: disqualifier <dev@disqualifier.me>
