feat: add voice transcription as nanorepo skill (#326)

Add voice transcription skill package at
.claude/skills/add-voice-transcription/ so it can be applied via the
skills engine. Skill adds src/transcription.ts (OpenAI Whisper), modifies
whatsapp.ts to detect/transcribe voice notes, and includes intent files,
3 test cases, and 8 skill validation tests.

Also fixes skills engine runNpmInstall() to use --legacy-peer-deps,
needed for any skill adding deps with Zod v3 peer requirements.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

.claude/skills/add-voice-transcription/modify/src/channels/whatsapp.ts.intent.md

@@ -0,0 +1,27 @@
+# Intent: src/channels/whatsapp.ts modifications
+
+## What changed
+Added voice message transcription support. When a WhatsApp voice note (PTT audio) arrives, it is downloaded and transcribed via OpenAI Whisper before being stored as message content.
+
+## Key sections
+
+### Imports (top of file)
+- Added: `isVoiceMessage`, `transcribeAudioMessage` from `../transcription.js`
+
+### messages.upsert handler (inside connectInternal)
+- Added: `let finalContent = content` variable to allow voice transcription to override text content
+- Added: `isVoiceMessage(msg)` check after content extraction
+- Added: try/catch block calling `transcribeAudioMessage(msg, this.sock)`
+  - Success: `finalContent = '[Voice: <transcript>]'`
+  - Null result: `finalContent = '[Voice Message - transcription unavailable]'`
+  - Error: `finalContent = '[Voice Message - transcription failed]'`
+- Changed: `this.opts.onMessage()` call uses `finalContent` instead of `content`
+
+## Invariants (must-keep)
+- All existing message handling (conversation, extendedTextMessage, imageMessage, videoMessage) unchanged
+- Connection lifecycle (connect, reconnect, disconnect) unchanged
+- LID translation logic unchanged
+- Outgoing message queue unchanged
+- Group metadata sync unchanged
+- sendMessage prefix logic unchanged
+- setTyping, ownsJid, isConnected — all unchanged