speaklisted

# Spoken-reply control Controls whether Claude Code reads its replies aloud. This is the *output* side; Claude Code's built-in `/voice` is the *input* side (dictation). The user ran `/speak $ARGUMENTS`. The command has already executed: !`"${CLAUDE_PLUGIN_ROOT}/bin/claude-speak" $ARGUMENTS` ## What to do Report the result in **one short line**. Examples of the whole reply: - `Voice on — af_heart at 1.0x.` - `Voice off.` - `Speed now 1.2x.` - `3 replies waiting from api-server and GitHub.` - `Reading README.md — about 4 minutes.` Rules: - No preamble, no explanation, no code blocks, no bullet lists, no follow-up offers. - If the output above is empty, an error, or still contains a literal `$ARGUMENTS`, run `"${CLAUDE_PLUGIN_ROOT}/bin/claude-speak" $ARGUMENTS` yourself with Bash and report that result instead. - If the output says to run `claude-speak install`, say so in one line — the neural voice model has not been downloaded yet. - If the argument was not recognised, say so in one line and list the valid ones inline: `on, off, stop, status, speed <n>, voice <name>, max <n>, mode <queue|interrupt|drop>, guard <on|off|test>, sound <file|off|default>, queue, hold <on|off>, pending, play [all|project], clear [all|project], read <file>, voices, audition, install, restart, log`. Reference — settings live in `~/.config/claude-speak/config.json`, the model and held replies in `~/.local/share/claude-speak/`, and the daemon is the systemd user service `claude-speak.