| name | respond-with-voice |
| description | Reply to the user with a voice note instead of (or in addition to) text.
Use when: (a) the incoming user message had `voice_in: true` meta (mirror modality),
(b) the user explicitly asked for a voice reply, or (c) the reply is short plain
language that is voice-friendly. DO NOT use when the reply contains code blocks,
tables, URLs the user needs to copy, or anything longer than ~3 sentences.
|
| allowed-tools | mcp__voice-tts__synthesize, mcp__telegram__reply_voice |
respond-with-voice
When invoked, you have a reply_text you want to deliver as audio.
-
Call mcp__voice-tts__synthesize with text=<reply_text>. This returns
{audio_path, duration_seconds, voice_used}.
-
Call mcp__telegram__reply_voice (provided by the telegram channel plugin)
with audio_path=<the path from step 1>. The channel uploads it as a
Telegram voice note in the current chat.
-
If the reply ALSO contains a code block, a long table, or a URL the user
needs to copy/click, additionally send those parts as text via the channel's
normal text reply so the user has copy-pasteable material.
-
Confirm in plain language (one short sentence) that you've sent the voice.
Do NOT also TTS this confirmation — it would create an echo loop.
When not to call this skill
- Replies containing more than ~3 sentences of dense info — use text
- Code blocks, JSON, tables — use text
- Error messages with stack traces — use text
- When the user has explicitly disabled voice in this thread (check
/admin/conversations for thread preferences)
