## Web Search & Fetch (Ghostfetch)
- **Skill:** `ghostfetch` (ClawHub)
- **Binary:** `~/.openclaw/workspace/tools/ghostfetch`
- **Description:** CLI web search and page fetcher — single binary, no browser required

**Usage:**
```bash
# Search the web (DuckDuckGo default)
ghostfetch "your search query"
ghostfetch "query" -e brave       # Use Brave
ghostfetch "query" -e google      # Use Google
ghostfetch "query" -n 5           # Limit to 5 results

# Fetch pages as markdown
ghostfetch fetch https://example.com -m
ghostfetch fetch url1 url2 -m -p 2  # Parallel fetch

# Extract links
ghostfetch links https://example.com
ghostfetch links https://example.com -f "github"  # Filter by pattern
```

**Note:** Always use `-m` (markdown mode) when reading page content — extracts main content and saves tokens vs raw HTML.

---

## TTS (Piper - Free Local)
- **Provider:** Piper (local, free)
- **Skill:** `local-piper-tts-multilang-secure` (ClawHub)
- **Default Voice:** Alan (en_GB, British male)
- **Available Voices:** Ryan (US male), Amy (US female), Alan (UK male)
- **Location:** `~/.openclaw/workspace/skills/local-piper-tts-multilang-secure/`

**Usage:**
```javascript
// Via skill
const skill = require('./index.js');
skill.tts({ text: 'Hello Anthony!', voice: 'en_GB-alan-medium' });
```

**Config file:** `~/.openclaw/workspace/skills/local-piper-tts-multilang-secure/config.json`

**Note:** Piper replaced ElevenLabs (cost savings). Fully offline, no API keys needed.

## Home Assistant

- **URL:** http://homeassistant.kangaroo-eel.ts.net:8123
- **Config:** ~/.config/home-assistant/config.json
- **Entities:** 225 total (16 lights, 29 switches, 21 media players, 2 vacuums, 2 fans, 3 scenes, 1 cover)

## Transcription (Groq Whisper - DEFAULT)

- **Skill:** `groq-whisper`
- **Location:** `~/.openclaw/workspace/skills/groq-whisper/`
- **Command:** `./scripts/transcribe.sh <audio-file> [flags]`
- **Model:** `whisper-large-v3` (fast & free via Groq)
- **Uses:** `GROQ_API_KEY` (already configured)

**Usage:**
```bash
# Basic transcription
transcribe.sh audio.m4a

# With options
transcribe.sh audio.ogg --language en --out result.txt --json
```

**Note:** Groq Whisper is free and fast. Use this instead of OpenAI Whisper.

### Downstairs
- **Name:** Downstairs roborock
- **Device ID:** 5HwaUBFem0YqXvijnABoCq

### Upstairs
- **Name:** Upstairs roborock  
- **Device ID:** 1dppdZmcbh6dSWbg6NPOZ5

### Quick Commands
```bash
# Check status
roborock status --device_id "5HwaUBFem0YqXvijnABoCq"

# Start cleaning
roborock command --device_id "5HwaUBFem0YqXvijnABoCq" start

# Stop and go home
roborock command --device_id "5HwaUBFem0YqXvijnABoCq" home

# Check consumables
roborock consumables --device_id "5HwaUBFem0YqXvijnABoCq"
```