Initial commit: workspace setup with skills, memory, config

skills/kimi-tts-custom/SKILL.md

@@ -0,0 +1,104 @@
+---
+name: kimi-tts-custom
+description: Custom TTS handler for Kimi that generates voice messages with custom filenames (Kimi-XXX.ogg) and optionally suppresses text output. Use when user wants voice-only responses with branded filenames instead of default OpenClaw TTS behavior.
+---
+
+# Kimi TTS Custom
+
+## Overview
+
+Custom TTS wrapper for local Kokoro that:
+- Generates voice with custom filenames (Kimi-XXX.ogg)
+- Can send voice-only (no text transcript)
+- Uses local Kokoro TTS at 10.0.0.228:8880
+
+## When to Use
+
+- User wants voice responses with "Kimi-" prefixed filenames
+- User wants voice-only mode (no text displayed)
+- Default TTS behavior needs customization
+
+## Voice-Only Mode
+
+**⚠️ CRITICAL: Generation ≠ Delivery**
+
+Simply generating a voice file does NOT send it. You must use proper delivery method:
+
+### Correct Way: Use voice_reply.py
+```bash
+python3 /root/.openclaw/workspace/skills/kimi-tts-custom/scripts/voice_reply.py "1544075739" "Your message here"
+```
+
+This script:
+1. Generates voice file with Kimi-XXX.ogg filename
+2. Sends via Telegram API immediately
+3. Cleans up temp file
+
+### Wrong Way: Text Reference
+❌ Do NOT do this:
+```
+[Voice message attached: Kimi-20260205-185016.ogg]
+```
+This does not attach the actual audio file — user receives no voice message.
+
+### Alternative: Manual Send (if needed)
+If you already generated the file:
+```bash
+# Use OpenClaw CLI
+openclaw message send --channel telegram --target 1544075739 --media /path/to/Kimi-XXX.ogg
+```
+
+## Configuration
+
+Set in `messages.tts.custom`:
+```json
+{
+  "messages": {
+    "tts": {
+      "custom": {
+        "enabled": true,
+        "voiceOnly": true,
+        "filenamePrefix": "Kimi",
+        "kokoroUrl": "http://10.0.0.228:8880/v1/audio/speech",
+        "voice": "af_bella"
+      }
+    }
+  }
+}
+```
+
+## Scripts
+
+### scripts/generate_voice.py
+Generates voice file with custom filename and returns path for sending.
+
+**⚠️ Note**: This only creates the file. Does NOT send to Telegram.
+
+Usage:
+```bash
+python3 generate_voice.py "Text to speak" [--voice af_bella] [--output-dir /tmp]
+```
+
+Returns: JSON with `filepath`, `filename`, `duration`
+
+### scripts/voice_reply.py (RECOMMENDED)
+Combined script: generates voice + sends via Telegram in one command.
+
+**This is the correct way to send voice replies.**
+
+Usage:
+```bash
+python3 voice_reply.py "1544075739" "Your message here" [--voice af_bella]
+```
+
+This generates the voice file and sends it immediately (voice-only, no text).
+
+## Key Rule
+
+| Task | Use |
+|------|-----|
+| Generate voice file only | `generate_voice.py` |
+| Send voice reply to user | `voice_reply.py` |
+| Text reference to file | ❌ Does NOT work |
+
+**Remember**: Generation and delivery are separate steps. Use `voice_reply.py` for complete voice reply workflow.

skills/kimi-tts-custom/scripts/generate_voice.py

@@ -0,0 +1,86 @@
+#!/usr/bin/env python3
+"""
+Generate voice with custom Kimi-XXX filename using local Kokoro TTS
+Usage: generate_voice.py "Text to speak" [--voice af_bella] [--output-dir /tmp] [--speed 1.3]
+"""
+
+import argparse
+import json
+import os
+import sys
+import tempfile
+import urllib.request
+from datetime import datetime
+
+def generate_voice(text, voice="af_bella", output_dir="/tmp", model="tts-1", speed=1.3):
+    """Generate voice file with Kimi-XXX filename"""
+    
+    # Generate unique filename: Kimi-YYYYMMDD-HHMMSS.ogg
+    timestamp = datetime.now().strftime("%Y%m%d-%H%M%S")
+    filename = f"Kimi-{timestamp}.ogg"
+    filepath = os.path.join(output_dir, filename)
+    
+    # Call local Kokoro TTS
+    tts_url = "http://10.0.0.228:8880/v1/audio/speech"
+    
+    data = json.dumps({
+        "model": model,
+        "input": text,
+        "voice": voice,
+        "speed": speed
+    }).encode()
+    
+    req = urllib.request.Request(
+        tts_url,
+        data=data,
+        headers={"Content-Type": "application/json"}
+    )
+    
+    try:
+        with urllib.request.urlopen(req) as response:
+            audio_data = response.read()
+            
+        # Save to file
+        with open(filepath, "wb") as f:
+            f.write(audio_data)
+        
+        # Estimate duration (rough: ~150 chars per minute at normal speed, adjusted for speed)
+        estimated_duration = max(1, len(text) / 150 * 60 / speed)
+        
+        result = {
+            "filepath": filepath,
+            "filename": filename,
+            "size_bytes": len(audio_data),
+            "estimated_duration_seconds": round(estimated_duration, 1),
+            "voice": voice,
+            "speed": speed,
+            "text": text
+        }
+        
+        print(json.dumps(result))
+        return result
+        
+    except Exception as e:
+        error_result = {
+            "error": str(e),
+            "filepath": None,
+            "filename": None
+        }
+        print(json.dumps(error_result), file=sys.stderr)
+        sys.exit(1)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Generate voice with Kimi-XXX filename")
+    parser.add_argument("text", help="Text to convert to speech")
+    parser.add_argument("--voice", default="af_bella",
+                       help="Voice ID (default: af_bella)")
+    parser.add_argument("--output-dir", default="/tmp",
+                       help="Output directory (default: /tmp)")
+    parser.add_argument("--model", default="tts-1",
+                       help="TTS model (default: tts-1)")
+    parser.add_argument("--speed", type=float, default=1.3,
+                       help="Speech speed multiplier (default: 1.3)")
+    
+    args = parser.parse_args()
+    
+    generate_voice(args.text, args.voice, args.output_dir, args.model, args.speed)

skills/kimi-tts-custom/scripts/voice_reply.py

@@ -0,0 +1,119 @@
+#!/usr/bin/env python3
+"""
+Generate voice with Kimi-XXX filename and send via Telegram (voice-only, no text)
+Usage: voice_reply.py <chat_id> "Text to speak" [--voice af_bella] [--speed 1.3] [--bot-token TOKEN]
+"""
+
+import argparse
+import json
+import os
+import sys
+import subprocess
+import tempfile
+import urllib.request
+from datetime import datetime
+
+def generate_voice(text, voice="af_bella", output_dir="/tmp", model="tts-1", speed=1.3):
+    """Generate voice file with Kimi-XXX filename"""
+    
+    # Generate unique filename: Kimi-YYYYMMDD-HHMMSS.ogg
+    timestamp = datetime.now().strftime("%Y%m%d-%H%M%S")
+    filename = f"Kimi-{timestamp}.ogg"
+    filepath = os.path.join(output_dir, filename)
+    
+    # Call local Kokoro TTS
+    tts_url = "http://10.0.0.228:8880/v1/audio/speech"
+    
+    data = json.dumps({
+        "model": model,
+        "input": text,
+        "voice": voice,
+        "speed": speed
+    }).encode()
+    
+    req = urllib.request.Request(
+        tts_url,
+        data=data,
+        headers={"Content-Type": "application/json"}
+    )
+    
+    try:
+        with urllib.request.urlopen(req) as response:
+            audio_data = response.read()
+            
+        with open(filepath, "wb") as f:
+            f.write(audio_data)
+        
+        return filepath, filename
+        
+    except Exception as e:
+        print(f"Error generating voice: {e}", file=sys.stderr)
+        sys.exit(1)
+
+def send_voice_telegram(chat_id, audio_path, bot_token=None):
+    """Send voice message via Telegram"""
+    
+    # Get bot token from env or config
+    if not bot_token:
+        bot_token = os.environ.get("TELEGRAM_BOT_TOKEN")
+    
+    if not bot_token:
+        # Try to get from openclaw config
+        try:
+            result = subprocess.run(
+                ["openclaw", "config", "get", "channels.telegram.botToken"],
+                capture_output=True, text=True
+            )
+            bot_token = result.stdout.strip()
+        except:
+            pass
+    
+    if not bot_token:
+        print("Error: No bot token found. Set TELEGRAM_BOT_TOKEN or provide --bot-token", file=sys.stderr)
+        sys.exit(1)
+    
+    # Use openclaw CLI to send
+    cmd = [
+        "openclaw", "message", "send",
+        "--channel", "telegram",
+        "--target", chat_id,
+        "--media", audio_path
+    ]
+    
+    try:
+        result = subprocess.run(cmd, capture_output=True, text=True)
+        if result.returncode == 0:
+            print(f"✅ Voice sent successfully to {chat_id}")
+            return True
+        else:
+            print(f"Error sending voice: {result.stderr}", file=sys.stderr)
+            return False
+    except Exception as e:
+        print(f"Error: {e}", file=sys.stderr)
+        return False
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Generate and send voice-only reply")
+    parser.add_argument("chat_id", help="Telegram chat ID to send to")
+    parser.add_argument("text", help="Text to convert to speech")
+    parser.add_argument("--voice", default="af_bella", help="Voice ID (default: af_bella)")
+    parser.add_argument("--speed", type=float, default=1.3, help="Speech speed multiplier (default: 1.3)")
+    parser.add_argument("--bot-token", help="Telegram bot token (or set TELEGRAM_BOT_TOKEN)")
+    parser.add_argument("--keep-file", action="store_true", help="Don't delete temp file after sending")
+    
+    args = parser.parse_args()
+    
+    print(f"Generating voice for: {args.text[:50]}...")
+    filepath, filename = generate_voice(args.text, args.voice, speed=args.speed)
+    print(f"Generated: {filename}")
+    
+    print(f"Sending to {args.chat_id}...")
+    success = send_voice_telegram(args.chat_id, filepath, args.bot_token)
+    
+    if success and not args.keep_file:
+        os.remove(filepath)
+        print(f"Cleaned up temp file")
+    elif success:
+        print(f"Kept file at: {filepath}")
+    
+    sys.exit(0 if success else 1)